Download File

Home , Hindgut fermentation, Megatheriidae

TESTING FOUNDATIONAL TENETS OF STABLE ISOTOPE ECOLOGY ANALYSES IN

NEOTROPICAL MAMMALIAN COMMUNITIES, AND IMPLICATIONS FOR

TERRESTRIAL PALEOECOLOGY

Julia Victoria Tejada Lara

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy under the Executive Committee of the Graduate School of Arts and Sciences

COLUMBIA UNIVERSITY

2020

Julia Victoria Tejada Lara

Abstract

TESTING FOUNDATIONAL TENETS OF STABLE ISOTOPE ECOLOGY ANALYSES IN

NEOTROPICAL MAMMALIAN COMMUNITIES, AND IMPLICATIONS FOR

TERRESTRIAL PALEOECOLOGY

Julia V. Tejada Lara

Stable isotope analyses are powerful tools for reconstructing ancient ecologies and ecosystems, as they are independent of morphology and directly reflect dietary ecology. The application of stable isotope analyses, however, is not without limitations, as determination of food web dynamics using these methods often relies on poorly tested assumptions. The guiding thread of this thesis is the testing of foundational cornerstones on which these methods rely, in order to validate the suitability of applying these techniques to different mammalian clades, and to more reliably and confidently interpret the isotopic signals preserved in extinct organisms.

The first chapter of this thesis tests the validity of an important assumption behind the interpretation of stable carbon isotope analyses for understanding diet in terrestrial mammalian herbivores: if, as assumed for almost two decades, mammalian bioapatite d13C is enriched by

14‰ relative to dietary d13C. By analyzing new isotopic data from a never before assessed herbivorous group spanning a broad range of body masses—sloths (Xenarthra, Mammalia)— and other mammals with experimentally controlled or observationally known diets, I discovered

considerable variation in diet–bioapatite d13C enrichment among mammals. Statistical tests

(ordinary least squares, quantile, robust regressions, Akaike information criterion model tests) documented independence from phylogeny, and a previously unrecognized strong and significant correlation of d13C enrichment with body mass for all mammalian herbivores. A single-factor body mass model outperformed all other single-factor or more complex combinatorial models evaluated, including for physiological variables (metabolic rate and body temperature proxies), and indicated that body mass alone predicts d13C enrichment. These analyses, spanning more than 5 orders of magnitude of body sizes, yield a size-dependent prediction of isotopic enrichment across Mammalia and for distinct digestive physiologies, permitting reconstruction of foregut versus hindgut fermentation physiologies for fossils and refined mean annual paleoprecipitation estimates based on d13C of mammalian bioapatite.

Second, I sought to evaluate the existing paradigm governing identification of closed canopy rainforests in the fossil record using mammalian d13C data: the presence of mammals with dietary d13C <-31‰, which has only been observed in closed canopy rainforests in

Equatorial Africa, the only other tropical ecosystem sampled extensively. This chapter provides a

13 13 15 characterization of d Cbioapatite, d Chair and d Nhair of a modern mammalian community in western Amazonia, in Peru, to test if the isotopic structure of mammals in this Neotropical ecosystem is similar to those in African tropical rainforests. The results indicate that despite their

13 marked geographical and taxonomic differences, median d Cdiet values from closed canopy rainforests in Amazonia (-27.4‰) and equatorial Africa (-26.9‰) are not significantly different.

Amazonian mammals, however, seem to exploit a narrower spectrum of dietary resources than

13 equatorial African mammals, as depicted by the absence of highly negative d Cdiet values previously proposed as indicative of rainforests (<-31‰). I hypothesize that differential effects

of late Pleistocene extinction may be responsible for the ecological disparities among the two rainforests, by significantly reducing evolutionary time and dietary breadth reflected in the modern Amazonian mammalian community.

Finally, the third chapter of this dissertation evaluates assumptions behind d15N amino acid compound specific analyses in order to test the controversial hypothesis of carnivory and consumption of proteins of animal origin in fossil sloths. This analytical technique relies on three

15 15 main assumptions. First, that the offset between the d N of glutamic acid (d NGlx) and

15 phenylalanine (d NPhe) in the organism under study will increase with increasing trophic level.

15 15 Second, that the offset between d NGlx and d NPhe at the base of the food chain is relatively constant and has a value of -8.4‰ for C3 ecosystems. Third, that the trophic discrimination

15 15 factor in all ecosystems (the difference in d NGlx relative to d NPhe with increasing trophic level) is 7.6‰. The results of my experiments conducted on extant xenarthrans (sloths and anteaters) with controlled diets document that only the first assumption holds true. Rather than relying on an equation with constants introducing uncertainties and that are not applicable to

15 organisms feeding on a combination of items of different origin (e.g., C3 + C4 plants), d NGlx

15 and d NPhe values by themselves can accurately reconstruct the trophic position of organisms.

15 15 Indeed, the results on d NGlx and d NPhe herein obtained for five xenarthran species in controlled feeding experiments, combined with mammalian data available from the literature, show strong and significant correlations between these two AAs and with trophic positions. Both

15 15 the TP equation and the regression analyses of d NGlx and d NPhe suggest that the Pleistocene fossil ground sloths Mylodon darwinii and Nothrotheriops shastensis were not pure herbivores as commonly presumed, but rather that they were both mixed feeders/omnivores, incorporating items of animal origin in their diets.

Table of Contents

List of Figures ...... iv

List of Tables ...... vi

Acknowledgments ...... vii

Dedication ...... viii

Introduction ...... 1

Chapter 1: Body Mass Predicts Isotope Enrichment in Herbivorous Mammals ...... 7

1.1 Testing Diet-Dentition Enrichment Using Sloths, a Non-Model Group of Mammalian

Herbivores ...... 9

1.2 Diet-bioapatite d13C enrichment in sloths and other herbivorous mammals ...... 10

1.3 Body mass explains high variation of e*diet-bioapatite across mammals ...... 13

1.4 Significance of implementing body mass as predictor of e*diet-bioapatite ...... 25

Chapter 2: Comparative Isotope Ecology of Rainforest Mammals: Amazonia Alters Current

Paradigms ...... 28

2.1 South American and African mammalian communities ...... 30

2.1 Sampling localities and d13C of dietary sources of Amazonian mammals ...... 31

2.2 Materials and Methods ...... 32

13 2.3 d Cdiet of western Amazonian mammals ...... 34

15 13 2.4 d Nhair and hair-bioapatite d C enrichment ...... 39

13 2.5 Comparisons of d Cdiet of mammalian herbivore communities from western Amazonia

and equatorial Africa ...... 40

2.6 d13C enrichment between proteinaceous tissues and bioapatite ...... 45

2.7 Potential influence of Pleistocene extinction on niche occupation of Amazonian mammals

...... 49

Chapter 3: TESTING DIETARY-CONSUMER d15N OF AMINO ACIDS IN MODERN

XENARTHRAN SPECIES AS A BASELINE FOR INTERPRETATION OF FOSSIL SLOTH

AMINO ACID COMPOUND SPECIFIC DATA ...... 55

3.1 Animal and diet samples ...... 60

3.2 Amino acid d15N from hair keratin and diets ...... 61

3.3 Nitrogen isotope analyses of amino acids ...... 62

3.4 Trophic position estimation and testing ...... 63

3.5 Bulk d15N of hair keratin and diets of modern species ...... 63

3.6 Bulk d15N of hair keratin for the fossil species ...... 67

3.7 AACSIA d15N of hair keratin and diet samples ...... 69

3.8 d15N of glutamic acid and phenylalanine ...... 76

3.9 Trophic position estimations for modern and fossil xenarthrans using constant b and TDF

values ...... 78

3.10 Assumptions behind the equation to calculate trophic positions using AA d15N values .. 79

3.11 Trophic position estimations ...... 83

15 15 3.12 Correlation between d NGlx and d NPhe ...... 84

3.13 Comparison of d15N AACSIA results to interpretations from other proxies ...... 87

3.14 Conclusions ...... 88

Epilogue ...... 90

References ...... 94

Appendix A ...... 111

Appendix B ...... 127

iii

List of Figures

CHAPTER 1 Figure 1.1 Mammalian phylogenetic tree for sampled herbivores, mapped in a space defined

13 by d C diet-bioapatite enrichment (e*diet-bioapatite)

Figure 1.2 Correlations between body mass and e*diet-bioapatite in controlled and constrained diet studies

Figure 1.3 Correlations between e*diet-bioapatite and different life history traits

CHAPTER 2

13 Figure 2.1 d Cdiet (calculated from dental bioapatite) for all western Amazonian mammals analyzed

13 Figure 2.2 d Cdiet values and distributions of the herbivore mammalian communities in western Amazonia and equatorial Africa

15 13 15 13 13 Figure 2.3 d Nhair vs d C e*bioapatite-keratin and Nhair vs d C e*diet-keratin (d Cdiet from bioapatite) of Amazonian mammals

13 Figure 2.4 d Cdiet values for both western Amazonian and equatorial African mammals plotted into their respective phylogenies

CHAPTER 3 Figure 3.1 Bulk d15N of the hair samples for the five modern xenarthran species and the two fossil sloths sampled Figure 3.2 Serial sampling of the hair for the fossil ground sloths Mylodon darwinii and Nothrotheriops shastensis Figure 3.3 Median d15N values for individual AA of Bradypus’ hair keratin (red) and its monospecific diet Ficus elastica Figure 3.4 Median d15N values for individual AA of Choloepus’ hair keratin (red) and its diet Figure 3.5 Mean d15N values for each AA for Cyclopes’ hair keratin and its mixed diet Figure 3.6 Mean d15N values for each AA for Myrmecophaga and Tamanadua’s hair keratin and diet iv

Figure 3.7 Mean AA d15N values for hair keratin of Nothrotheriops shastensis and Mylodon darwinii

15 15 Figure 3.8 Correlations between d NPhe and d NGlx for different mammalian species

List of Tables

CHAPTER 1 Table 1.1 Summary of isotopic analyses: d13C and e* for modern and extinct sloths Table 1.2 Summary of quantitative analyses

CHAPTER 2 Table 2.1 Summary of d13C results (‰) from dental bioapatite of western Amazonian mammals and equatorial Africa per taxonomic group

15 13 Table 2.2 Summary of results for d Nhair and d Chair (standardized to 1750 values) and

13 d C e*diet-keratin for modern western Amazonian mammals

CHAPTER 3 Table 3.1 Bulk d15N of the hair samples for the two specimens of Bradypus variegatus Table 3.2 Bulk d15N of the Bradypus’s diet samples (Ficus elastica, rubber plant) Table 3.3 Bulk d15N of the hair samples for the different specimens of Choloepus hoffmanii Table 3.4 Bulk d15N of Choloepus’ homogenized diet samples Table 3.5 Bulk d15N of hair and diet samples of Cyclopes didactylus Table 3.6 Bulk d15N of the hair and diet samples of Myrmecophaga tridactyla and Tamandua tetradactyla Table 3.7 Bulk d15N (‰ vs. AIR) and C:N ratio (mol/mol) for hair segments of Mylodon darwinii and Nothrotheriops shastensis Table 3.8 AA and bulk tissue δ15N values for modern and fossil xenarthran species Table 3.9 AA and bulk tissue δ15N values for dietary samples of the modern xenarthran species

15 15 15 15 Table 3.10 Summary of the b (d NGlx-d NPhe in the diets), TDF, and ∆d NGlx-d NPhe values for all species in this study

Acknowledgments

Dissertation Committee: John Flynn, Sidney Hemming, Ross MacPhee, Thure Cerling, Jin Meng

Institutions: Department of Earth and Environmental Sciences and GSAS, Columbia University in the City of New York, American Museum of Natural History, Zoologico de Huachipa (Lima,

Peru), Museo de Historia Natural (Lima, Peru), Florida Museum of Natural History, SIRFER-

University of Utah, University of Hawaii at Manoa, Yale Peabody Museum, Museo de la Plata

(La Plata, Argentina), National Museum of Natural History, Natural History Museum of Los

Angeles County.

Funding: NSF-Inter University Training in Continental Scale Ecology (1137336), NSF-PCP

PIRE (0966884), GSAS-Columbia University, American Museum of Natural History (Division of Paleontology Frick Fund and RGGS), Paleontological Society, Society of Vertebrate

Paleontology, The Society of Systematic Biologists, Chevron Student Initiative Fund (LDEO,

Columbia University)

Individuals (alphabetical order): Pierre-Olivier Antoine, Patrice Baby, Ana Balcarcel, Jason

Burke, Suvankar Chakraborty, John Denton, Neil Duncan, Jim Ehleringer, Maddie Foote, Alana

Gishlick, Tyler Huth, IsoCamp Instructors, Gloria Lara, Verne Lee, Bruce MacFadden, Christy

Mancuso, Kaleigh Matthews, Carl Mehling, Nicole Mihnovetz, Carol Mountain, Tamsin

O’Connell, Sally Odland, Brian Popp, Anna Ragni, Rodolfo Salas-Gismondi, Polly Scarvalone,

Kelly Speer, Juan Tejada, Juan F. Tejada, Ivan Tejada, Sophie Tejada, Kevin Uno, Mario

Urbina, Natalie Wallsgrove, Yasmin Yabyabin… and so many others!

vii

Dedication

To Rodolfo Salas-Gismondi,

For introducing me to the world of

paleontology with his contagious passion

viii

Introduction

American fossil mammal communities.

Stable isotope analyses are powerful tools for reconstructing ancient ecologies and ecosystems, as they are independent of morphology and can directly reflect dietary ecology.

These techniques rely on two underlying premises. First, that isotopic values of any given chemical element differ across diverse pools and reservoirs in the biosphere, and therefore that different foods and waters are isotopically varied. Second, because an individual’s body tissues are synthesized from the food and water they ingest, their body tissues are isotopically linked in a predictable way to their food/water intake. Thus, stable isotope analyses offer a rare route to document the consumed diet itself, rather than inferences based on indirect morphological indicators, representing a particularly useful and potentially more reliable method to be applied to fossil samples.

The application of stable isotope analyses, however, is not without caveats, as determination of food web dynamics using these methods relies on some poorly tested assumptions. The guiding thread of this thesis is the testing of foundational cornerstones on which these methods rely, in order to validate the appropriateness of applying these techniques to different mammalian clades, and to more reliably and confidently interpret the isotopic signals preserved in extinct terrestrial organisms.

The first chapter of this thesis evaluates an important assumption behind the interpretation of stable carbon isotope analyses in mammalian herbivores: that the signatures recorded in tissues accurately reflect the ingested diet. Indeed, carbon isotopic signatures recorded in vertebrate tissues derive from ingested food and thus do reflect ecologies and ecosystems. For almost two decades, however, most carbon isotope-based ecological interpretations of extant and extinct herbivorous mammals have used a single diet–bioapatite enrichment value (14‰) applied to all taxa (e.g., (1)). Recognition of the body mass dependence of carbon isotope enrichment in mammals is essential for accurate understanding of the relative d13C differences among members of a herbivore community, which are used to reconstruct feeding niches and environments, and to evaluate ecological and environmental changes through time. For instance, species with similar d13C values, which would be interpreted as overlapping in resource usage when using a single standard diet-bioapatite enrichment value, could reveal distinct niche partitioning if body mass is taken into account. Similarly, imprecise or even misleading reconstructions of dietary changes through time can arise if the d13C of the plants consumed have been inferred with a single enrichment value for all herbivores. For instance, application of my proposed body mass-corrected diet-bioapatite enrichment could refine determination of the precise time when hominins began incorporating C4 plants in their diets and

2 the proportion that these represented in the evolution of hominin diets over time. Indeed, assuming that this single value applies to all herbivorous mammals, from tiny monkeys to giant elephants, overlooks potential effects of distinct physiological and metabolic processes on carbon fractionation that might yield different enrichment patters across species. This study is built on a few other studies that identified enrichment pattern differences between diet intake and tissues in some mammalian herbivore clades( e.g. (2–5)). The concern of corroborating the validity of this diet-bioapatite enrichment value was particularly warranted because I aimed to evaluate the feeding ecology of fossil sloths. Sloths (Folivora: Xenarthra) represent one of the least studied groups of mammals, although they are crucial to understanding many pre-Pleistocene South

American mammal faunal communities given the taxonomic abundance and variety of habitats occupied by this group over the past 35 Myr. Prior interpretations of extinct sloth ecologies relied predominantly on morphology. Consequently, I aimed to test those interpretations by means of stable isotope analyses, to provide ecological reconstructions independent of morphology. However, because sloths are outliers in a number of life-history traits characteristic of other mammals (6,7), the standard ungulate-based assumption of a 14‰ diet–bioapatite enrichment (8) is tenuous until directly tested. This study evaluated d13C diet–bioapatite enrichment for herbivores across a wide diversity of mammal clades (including modern sloths, a herbivore group never before assessed). Because modern sloths vastly underrepresent the ecological and morphological diversity the group had until as late as 10 000 years ago, proper representation of the group in isotopic analyses demands inclusion of extinct representatives.

Thus, I analyzed dental bioapatite of the Pleistocene ground sloth Mylodon darwinii (Mylodon

Cave, Chilean Patagonia).

The second chapter of this dissertation evaluates the way carbon stable isotope analyses are used to identify closed canopy rainforests in the fossil record. Closed canopy rainforests are important for climate (influencing atmospheric circulation, albedo, carbon storage, etc.) and ecology (harboring the highest biodiversity of continental regions). Of all rainforests, Amazonia is the world’s most diverse, including the highest mammalian species richness. But little is known about niche structure, ecological roles, and food resource partitioning of Amazonian

13 13 15 mammalian communities over time. Through analyses of d Cbioapatite, d Chair and d Nhair, I isotopically characterized aspects of feeding ecology in a modern western Amazonian mammalian community in Peru, to create the first comprehensive sampling of this ecosystem, further serving as a baseline for understanding the evolution of Neotropical rainforest ecosystems. By comparing these results with data from equatorial Africa, the only broadly sampled mammalian communities from closed canopy rainforests, which to date has served as the de facto baseline for all later studies (1,9,10), I evaluated the potential influences of distinct phylogenetic and biogeographic histories on the isotopic niches occupied by mammals in analogous tropical ecosystems. Indeed, the unique presence of mammalian herbivores with

13 d Cdiet values <-30‰ in equatorial Africa led to hypothesize that this is a defining trait for closed canopy rainforest identification using mammalian d13C data. By comparing the isotopic structure of these two mammalian communities, equatorial Africa and Western Amazonia, this study aimed to define the d13C range for closed canopy rainforests across continents serving as a baseline for understanding changes in the Amazonian ecosystem through time. Moreover, evaluation of resource exploitation within and among mammalian clades across both ecosystems provides a new perspective on the ecological complexities of tropical South American mammals.

Finally, the third chapter of this dissertation involves the novel application of d15N analyses (11–16) on specific amino acids to test a controversial hypothesis of carnivory and consumption of proteins of animal origin on fossil sloth species. Traditionally, bulk nitrogen stable isotope analyses (d15N) have been used to evaluate trophic interactions due to an often- found increase of 3-5‰ in d15N per trophic level change (17,18). However, substantial variation in d15N values has also been observed among trophic guilds (18,19), showing that the use of d15N alone can be a misleading proxy to identify trophic levels. Indeed, interpretation of bulk d15N values often fail to acknowledge variations in ecosystems’ baseline d15N values which can greatly differ depending on the way primary producers synthesize nitrogen and the isotopically

- different types of inorganic nitrogen available (e.g., N2, NH3, NO 3, (20)).

Studies relating population densities and body mass suggested that pre-Pleistocene South

American mammalian communities were imbalanced because the biomass of herbivores significantly surpassed the energetic requirements of carnivore (21). Likewise, it has been suggested that fossil ecosystems could not have supported the plant biomass demands of all of the presumed large herbivores, and as a consequence there must be secondary consumers disguised as herbivores in the South American fossil record (21). Because of their simple (and potentially functionally versatile) dentition, fossil sloths have been identified as the most likely putative herbivore candidates to have instead occupied scavenger niches (22). Fossil sloths are generally considered herbivores (23,24), like their modern relatives, but their extremely high diversity (in terms of body weight [4-4000 kg], geographical distribution [from Patagonia to

Alaska], habitats [coast, mountains, rainforest] and inferred locomotion [terrestrial, fossorial, arboreal, semiarboreal, and semiaquatic]) suggests that these were ecologically more versatile than traditionally thought. Considering that 1 out of 3 fossil mammals in South America was a

5 xenarthran (the group including sloths, armadillos, and anteaters (25)), the discovery that fossil sloths could have had ecologies differing from those traditionally attributed to them would radically alter understanding of how the mammalian communities were structured in ancient

South American ecosystems. The technique of amino acid compound specific isotope analysis permits direct testing of this hypothesis (12,26,27). To confidently apply this technique, however, I first must test if the underlying assumptions behind the method hold true for modern sloths and other closely related species under different feeding regimes. For instance, this technique relies on the assumption that the isotopic differences among amino acids in primary producers, and the 15N-enrichment between a consumer and its food, are constant across all organisms (12,28). This third chapter, therefore, aims to determine relationships between diet and tissues (diet-amino acid d15N) in modern xenarthrans with different feeding regimes, generating essential foundational data for understanding diet-tissue chemical partitioning in modern mammals more broadly, and to then be able to reliably interpret d15N isotopic amino acid data from fossil species to understand their trophic relationships.

Chapter 1: Body Mass Predicts Isotope Enrichment in Herbivorous

Mammals

Julia V. Tejada-Lara1,2,3*, Bruce J. MacFadden4, Lizette Bermudez5, Gianmarco Rojas5, Rodolfo

Salas-Gismondi2,3,6, John J. Flynn2

1Department of Earth and Environmental Sciences, Lamont-Doherty Earth Observatory, Columbia University. New York. 2Division of Paleontology, American Museum of Natural History. New York. 3Departamento de Paleontología de Vertebrados, Museo de Historia Natural, Universidad Nacional Mayor de San Marcos. Lima, Perú. 4Florida Museum of Natural History, University of Florida. Gainesville, Florida. 5Zoológico de Huachipa. Lima, Perú. 6BioGeoCiencias Lab, Facultad de Ciencias y Filosofía/CIDIS, Universidad Peruana Cayetano Heredia, Lima, Perú

Tejada-Lara, Julia V., et al. "Body mass predicts isotope enrichment in herbivorous

mammals." Proceedings of the Royal Society B: Biological Sciences 285.1881 (2018):

20181020.

Understanding the structure and history of modern ecosystems and vertebrate communities requires detailed knowledge of extinct and extant animal feeding ecology. This information permits assessment of paleoclimate, vegetation structure, niche partitioning and predator-prey interactions in fossil ecosystems, so it is crucial to understanding habitat use and trophic specializations over time. Inferences of feeding ecology of extinct vertebrates, however, typically can be inferred only indirectly from morphological data (e.g., dental hypsodonty, enamel microwear and mesowear, cranial shape adaptations, etc.) constrained by correlations between diet and habitats in modern ecosystems. Stable isotope analysis is a useful tool to 7 reconstruct ancient ecologies and ecosystems, as it is independent from morphology and directly reflects dietary ecology. Because the carbon that animals use to synthesize different tissues relates directly to the ingested food (“you are what you eat”, (2), dietary isotopic signatures get recorded in organismal cells and tissues (1,3–5,8,17,29,30). However, during incorporation of carbon from diet to tissue, kinetic isotope effects of an array of physiological processes change the 13C/12C ratio (d13C) of ingested food to that ultimately expressed in tissues (i.e., isotope apparent fractionation, discrimination, or enrichment [e*]; (8)). For mammalian herbivores, the d13C of enamel bioapatite, the most frequently analyzed tissue in paleontology because of its lower susceptibility to alteration during fossilization (31), has been widely assumed to be enriched by ~14‰ with respect to dietary d13C values, based on a study of modern ungulates and their likely plant foods (8). This 14‰ value has been systematically but generally uncritically applied in virtually all subsequent dietary and environmental interpretations of fossil and extant mammalian herbivores, regardless of body size, phylogenetic affinities, or primary diet source.

Intriguingly, though, some studies identified enrichment pattern differences between diet intake and tissues in other mammalian herbivore clades (e.g., (2–5)). If these differences are consistent and significant across mammalian clades, size classes or dietary ecologies, they merit concern, as they imply that some previous interpretations of animal ecologies and reconstructed environments could be mistaken.

This study evaluates d13C diet-bioapatite enrichment for herbivores across a wide diversity of mammal clades (including non-ungulates and an herbivore group never before assessed— sloths). Isotope enrichment was determined through controlled feeding experiments or constrained diet analyses to test the prevailing assumption that a single enrichment pattern

8 holds for all plant-eating mammals, regardless of life history attributes or ecology. If it does not, we seek to find and test explanatory variables to predict this value.

1.1 Testing Diet-Dentition Enrichment Using Sloths, a Non-Model Group of

Mammalian Herbivores

As described in Darwin’s “Voyage of the Beagle”, fossil native mammals from South

America include the “strangest animals ever discovered” [11, pp. 80]. More than 170 years later, because of their lack of close modern relatives, ecological analogs, or lengthy ghost lineages, the phylogenetic relationships and ecologies of many of these extinct endemic mammal groups remain unresolved. One of the least studied groups of mammals are sloths (Folivora: Xenarthra), although they are crucial to understanding many pre-Pleistocene South American mammal faunal communities given the taxonomic abundance and variety of habitats occupied by this group over the past 35 million years. Prior interpretations of extinct sloth ecologies relied predominantly on morphology. Consequently, we aim to test those interpretations by means of stable isotope analyses, and provide ecological reconstructions independent of morphology.

However, because sloths are outliers in a number of life history traits characteristic of other mammals (6,7), the standard ungulate-based assumption of a 14‰ diet-bioapatite enrichment (8) is tenuous until directly tested. This caution is particularly warranted because sloths are foregut fermenters, have long food particle mean retention times, and produce more methane than other herbivores of similar body masses (the latter experimentally measured in Choloepus (7)). Some of these parameters have been suggested as potential drivers of the diet-bioapatite enrichment differences observed in mammals (5). Thus, determination of the d13C diet-bioapatite enrichment

[e*diet-bioapatite] in extant sloths provides a phylogenetically-independent test of ungulate-based results in another herbivorous group of mammals. It also becomes indispensable as a first step in

9 assessing isotopic values recorded in extinct sloth bioapatite, by creating a baseline for interpretation of analytical results from fossils.

1.2 Diet-bioapatite d13C enrichment in sloths and other herbivorous mammals

Bioapatite is of particular interest because it constitutes the inorganic matrix of bones and teeth, and thus is more likely to be preserved unaltered in the fossil record (especially that from tooth enamel) relative to organic tissues. In addition, the carbonate radical of bioapatite forms from blood bicarbonate, derived primarily from ingested carbohydrates, the main source of carbon in herbivorous diets (4,33). Bioapatite has proven to be a better predictor of the isotope value of the whole diet than are organic tissues (e.g., collagen, keratin), which mainly track the isotopic composition of the protein(s) used to synthesize them (3,4,33). Because sloths lack dental enamel, this study analyzes sloth dentine bioapatite relative to enamel bioapatite of other mammals. The d13C of enamel and dentine bioapatite is considered equivalent if a primary signal is preserved, because any bioapatite (bone, dentine, enamel) is in isotopic equilibrium with blood bicarbonate (34). In modern specimens, diagenetic alteration is not an issue. A potential problem when dealing with fossil samples is that during taphonomic processes, bone or dentine bioapatite can recrystallize and thus would be susceptible to isotope exchange in this process (35). For modern bone or dentine, although strong, lengthy laboratory treatments can cause recrystallization (e.g., using 1M acetic acid to remove secondary carbonates (35–38)), short treatments with weak 0.1M acetic acid (the protocol used herein) resulted in no detectable change between the d13C of enamel and ivory of elephants in protocol tests ((29), but see (38) for the effect of longer treatments).

We analyzed d13C values of dental bioapatite, diet, and fecal samples of extant sloths

(Bradypus variegatus and Choloepus hoffmanni) from the Huachipa Zoo (Lima, Peru) kept under

10 controlled and homogeneous feeding conditions (Table 1.1). Because modern sloths vastly underrepresent the ecological and morphological diversity the group had until as late as 10,000 years ago, proper representation of the group in isotopic analyses demands inclusion of extinct representatives. Thus, we analyzed dental bioapatite of three Pleistocene ground sloth specimens

(Mylodon darwinii [Mylodon Cave, Chilean Patagonia] and Nothrotheriops shastensis [Gypsum and Rampart Caves]), recovered in association with their dung (Table 1.1). Dung d13C is a direct indicator of dietary d13C, because there is minimal or no apparent fractionation between diet and feces (e*diet-feces) across mammals (30), as corroborated in our analyses of modern sloth feces

(table 1). Collagen atomic C:N ratio was used as a proxy to test for diagenetic alteration of the isotopic composition for Pleistocene sloths (this proxy assesses collagen quality and concentration, which tends to decrease and degrade with the age of the sample (39)). Of the three fossil samples, only Mylodon darwinii showed a C:N ratio corresponding to an unaltered sample

(C:N=3.3), and therefore it is the only specimen discussed further. Values of e*diet-bioapatite for the other mammals in this study, with either controlled or known diets and associated isotopic values, were gathered from the literature.

Our results (table 1) showed that the three-toed sloth (Bradypus variegatus, n=5) has an

13 13 average d Cbioapatite value of -18.65‰ ± 1, which compared to the d C of its diet at the zoo (-

28.67‰ ± 1.79), results in a e*diet-bioapatite value of 10.31‰ ± 1.03. The two-toed sloth (Choloepus hoffmanni, n=10) has an average d13Cbioapatite value of -14.08‰ ± 0.66, representing a 13C- enrichment of 12.62‰ ± 0.68 relative to the d13C of its diet (-26.37‰). The d13Cbioapatite (-

12.46‰) of the extinct ground sloth Mylodon darwinii is on average 15.63‰ ± 0.51 (15.12‰ to

16.13‰) 13C-enriched relative to its diet determined from feces (-27.66‰ ± 0.49). The high 13C-

11 enrichment of Mylodon’s bioapatite relative to its diet place this taxon at the uppermost extreme of the e*diet-bioapatite spectrum observed in mammals.

Table 1.1: Summary of isotopic analyses: d13C and e* (diet, dental bioapatite, and feces) for sloths with known diets (through controlled-feeding [Bradypus, Choloepus] or constrained-diet feces analyses [†Mylodon]). 1Calculation based on a concentration-weighted linear mixing model (40). 2Assuming feces-diet enrichment similar to that of Bradypus.3Assuming feces-diet enrichment similar to that of Choloepus

Diet Dental bioapatite Feces 13 d Cdiet Contribution of each 13 13 13 TAXA Diet 13 d Cdiet d Cbioap d Cfeces e* component element [d C (‰)] to n e* (‰) n component (‰) (‰) (‰) (‰) (‰) final signature of diet1 Rubber plant Bradypus -28.67± -28.67± -18.65 ± 10.31 ± -28.21 0.48 ± (Ficus Monospecific diet 5 5 variegatus 1.79 1.79 1 1.03 ±0.53 0.54 elastica) Purina® -16.5 -1.50 DogChow ® Quinoa -25.8 -2.28 Broccoli -27.3 -3.75 Choloepus -26.371 -14.08 ± 12.62 ± -24.98 1.43 10 15 hoffmanni Sweet potato -27.9 -4.66 0.66 0.68 ±0.63 ±0.64 Carrot -25.3 -4.14 Spinach -28.1 -7.29 Rubber plant -29.4 -2.75 †Mylodon Calculated -27.172 to 15.122 to - 1 -12.46 1 -26.7‰ - darwinii from dung -28.143 16.133

Placed in a broader mammalian context, results for three sloth taxa span a range of e*diet- bioapatite variation (>5‰) never before seen in any mammalian clade (figure 1.1), encompassing almost the entire range observed across all mammalian herbivores. The range of e*diet-bioapatite variation in Artiodactyla and Perissodactyla is small (1.2‰ and 1.7‰, respectively). Values of e*diet-bioapatite in Artiodactyla range from 12.89‰ in pigs (Sus scrofa) to 14.65‰ in cattle (Bos taurus), whereas the range of variation observed in Perissodactyla spans from 13.2‰ in the zebra

(Equus burchelli) to 14.4‰ in the black rhinoceros (Diceros bicornis). The range of e*diet-bioapatite values (across 3.7‰) observed in Glires (i.e., rodents [Rodentia] and lagomorphs [Lagomorpha]) is the second largest after that of Xenarthra. The house mouse Mus musculus has the lowest e*diet-bioapatite value (9.1‰) of the spectrum of mammals analyzed, the prairie vole Microtus ochrogaster has a e*diet-bioapatite of 11.5‰ and the European rabbit Oryctolagus cuniculus has a 12 e*diet-bioapatite of 12.8‰. Three distantly related taxa, the giant panda Ailuropoda melanoleuca

(Carnivora), koala Phascolarctos cinereus (Diprotodontia: Marsupialia), and three-toed sloth

(Folivora: Xenarthra), cluster near the lower end of the observed e*diet-bioapatite spectrum with values ~10-10.5‰.

Figure 1.1: Mammalian phylogenetic tree for sampled herbivores, mapped in a space defined by d13C diet-bioapatite enrichment (e*diet-bioapatite). Colors represent clades (see ESM). Notice the extremely large range of e*diet-bioapatite variation in Folivora (sloths), spanning almost the entire spectrum of variation observed in Mammalia. Animal silhouettes are for reference and not to scale.

1.3 Body mass explains high variation of e*diet-bioapatite across mammals

Aiming to find explanatory variables correlated to the broad range of e*diet-bioapatite values now observed across mammals with controlled diets or well-known d13C values of diets in the wild (even among three members of only one clade, sloths), a series of physiologically-

13 correlated parameters were regressed against e*diet-bioapatite: body mass, phylogeny, basal metabolic rate, basal metabolic rate corrected for known covariance with body mass

(BMR/Kleiber value), average body temperature (rectal temperature), and breadth of body temperature variation (figure 1.2-1.3). Information for other parameters potentially relevant to understanding these e*diet-bioapatite differences (e.g., nutritional quality of diet, food particle mean retention time, CH4 yield, diversity and abundance of digestive tract microorganisms) were not available for most taxa with known values of e*diet-bioapatite.

2 Body mass shows a strong and significant positive correlation with e*diet-bioapatite (r =0.62

[all taxa included], p-value=0.0002, figure 1.2) across mammalian herbivores. Correlation between body mass and e*diet-bioapatite is even stronger if mammals are separated by digestive system type: foregut fermenters (r2=0.78, p-value=0.008) versus hindgut fermenters (r2=0.74, p- value=0.003) (figure 1.2). We tested for and rejected the possibility that the significance of the regression was driven by the effect of the outliers (particularly those of small size), by performing quantile and robust regression model analyses (table 1.2). Regression diagnostics identified the vole, panda, and koala as outliers, but only the vole influenced the regression

(Cook’s D>0.5). Its exclusion in a secondary analysis, however, did not significantly alter the coefficient values; in fact, it makes the relationship even stronger (table 1.2). Based on a reduced taxon dataset (i.e., excluding taxa with unknown values, such as Mylodon, Diceros, and Equus burchelli), we also performed generalized linear model analyses based on AICc tests, to assess whether a single variable, reduced model or various model subsets were better supported by the data than a complete global model that includes all the potential predictors (table 1.2). Three models were well supported (delta AICc <2), all three of which included body mass as the most

14 important predictor variable, with the model including only body mass selected as the best model

(table 1.2).

Figure 1.2: Correlations between body mass and e*diet-bioapatite in controlled and constrained diet studies. The regression formula across all mammalian herbivores (dotted line) should be applied when type of fermentation of the herbivorous mammal under study is unknown, or does not fall definitively within foregut or hindgut types of herbivore fermentation (e.g., giant panda). Formulae for foregut (greenish line) and hindgut (red line) fermenters should be applied when the type of digestive fermentation (foregut or hindgut) of the mammal under study is known or likely given its phylogenetic history (see ESM). Species included are: (1) Mus musculus, house mouse; (2) Microtus ochrogaster, prairie vole; (3) Oryctolagus cuniculus, European rabbit; (4) Choloepus hoffmanni, two-toed sloth; (5) Bradypus variegatus, three-toed sloth; (6) Phascolarctos cinereus, koala; (7) Sus scrofa, pig; (8) Ailuropoda melanoleuca, giant panda; (9) Lama guanicoe, guanaco; (10) Equus burchelli, zebra; (11) Equus caballus, horse; (12) Camelus bactrianus, Bactrian camel; (13) Bos taurus, cow; (14) Diceros bicornis, black rhinoceros; (15) Giraffa camelopardalis, giraffe; (16) Loxodonta africana, African elephant; (17) Mylodon darwinii, giant ground sloth (extinct). All values are in natural logarithmic scale. BM in kg.

Comparing different models of trait evolution, e*diet-bioapatite data best fitted a white noise model (lowest AICc, table S3), which assumes data come from a single normal distribution with no covariance structure among species, indicating no phylogenetic signal in the observed values of mammalian e*diet-bioapatite. No significant correlation was found between e*diet-bioapatite and basal metabolic rate (BMR) for all mammalian herbivores, nor for any analysis of BMR after its body mass component was removed (i.e., the Kleiber value, or the ¾ power to which BMR has been described to scale relative to body mass in mammals, (41)). There is a significant correlation of raw BMR for hindgut fermenters only, although the lack of significance for even hindgut fermenters in the BMR/Kleiber analysis indicates that body mass explains that correlation (figure

3 a-b). Although, no correlations were found between e*diet-bioapatite and breadth of rectal temperature variation, a strong significant correlation was identified between e*diet-bioapatite and average rectal temperature, but for foregut fermenters only (r2=0.89, p-value=0.01) and not for hindgut fermenters or all mammalian herbivores (figure 1.3 c-d). However, a regression analysis of average rectal temperature and body mass, reveals that this correlation in foregut fermenters also is not independent of body mass. Rates of metabolic biochemical reactions are controlled by temperature (42), but based on our dataset, analyses and tests, body temperature variables do not seem to play a role in mammalian patterns of carbon isotopic fractionation independently of body mass. Analytical experiments to propose and test specific mechanistic explanations for how body temperature, independent of or in association with factors driven by body mass, might affect e*diet-bioapatite is outside the scope of the present contribution, but deserve future exploration for a better understanding of e*diet-bioapatite variation in mammals. Overall, body mass is the best predictor of e*diet-bioapatite (lowest AICc, table 1.2), and it is the only parameter that significantly correlated with e*diet-bioapatite in all cases (i.e., when all taxa were included, as well as when

16 mammals were separated by digestive physiology into foregut and hindgut fermenters). This newly recognized body-size dependence of carbon isotope enrichment from ingested diet to bioapatite tissue formation requires application of size-calibrated e*diet-bioapatite calculations, rather than a single enrichment value across all mammalian herbivores (proposed formulae in figure

1.2), and potential reinterpretation of some prior dietary and ecological results dependent on a single ~14‰ value. Additional future controlled-feeding experiments across a large range of body sizes for both foregut and hindgut fermenters would permit refinement of the regression formulae for the body-mass e*diet-bioapatite correlations.

Table 1.2: Summary of quantitative analyses. (a) Summary of coefficients and significance values for linear regressions performed. Quantile regression estimates the conditional median of the response variable, unlike OLS that estimates the approximate conditional mean. Robust regression fittings (objective functions used were Huber and Tukey bisquare) are performed by iterated, re-weighted least squares analyses. The formulae proposed use the average of the coefficients given by the different regression models (see ESM). (b) Component models (delta AICc < 2) that best predict d13C diet-bioapatite enrichment (e*diet-bioapatite) in herbivorous mammals given current data. Body mass alone is the best predictor of enrichment. Other, more complex models that simultaneously assess more potential influences, do not significantly improve prediction of the enrichment in the AICc analyses. Abbreviations: df: degrees of freedom, logLik: maximum log-likelihood, AICc: Akaike Information Criterion corrected for small sample sizes, delta: difference in AICc between the current and the best model, Weight: prior weights used in model fitting.

A. Linear regressions intercept slope R2 p-value All taxa included OLS model fit 2.4 0.034 0.62 0.0002 OLS model fit excluding outliers 2.39 0.04 0.83 5.35 x 10-6 Quantile regression fit 2.37 0.041 Robust regression (Huber estimator) 2.4 0.036 Robust regression (Tukey bisquare estimator) 2.4 0.035 Average 2.39 0.037 Hindgut fermenters OLS model fit 2.42 0.032 0.74 0.003 OLS model fit excluding outliers 2.4 0.036 0.89 0.001 Quantile regression fit 2.48 0.022 Robust regression (Huber estimator) 2.43 0.03 Robust regression (Tukey bisquare estimator) 2.51 0.019 Average 2.45 0.028

Foregut fermenters OLS and robust model values 2.34 0.05 0.78 0.008 Quantile regression fit 2.24 0.07 Robust regression (Huber estimator) 2.34 0.05 Robust regression (Tukey bisquare estimator) 2.34 0.05 Average 2.34 0.06 B. Component models df logLik AICc delta AICc Weight Body mass alone 3 12.71 -17.02 0.00 0.52 Body mass + average rectal temperature 4 14.03 -15.62 1.41 0.26 Body mass + basal metabolic rate 4 13.91 -15.38 1.64 0.23

13 The d C enrichment that occurs between dietary input and metabolic CO2 (that supplies the dissolved inorganic carbon in blood, in isotopic equilibrium with bioapatite) has been attributed to digestive physiology (5), in particular methane (CH4) production (43). Methane,

13 C-depleted by almost 40‰ relative to the ingested diet, is a by-product gas during CO2 reduction (44), which in mammals is due to microbial activity during digestive processes (43).

As a consequence, and acknowledging that attribution of methane production as the principal driver of the e*diet-bioapatite differences observed in mammals remains only theoretical (but see (43) and (44)), one would expect herbivores producing large amounts of methane to also show large e*diet-bioapatite values, as has been experimentally demonstrated in cattle (5,43). The magnitude of microbial activity in the digestive tract, expressed as biochemical transformations leading to differential diet-tissue enrichment values (e*), is predicted to result from the interaction of variables such as the capacity of the fermentative vat, amount and diversity of microorganisms living in it, and metabolic activity of the microorganisms (45), as well as the particle mean retention time (7), and the quality and quantity of the host’s diet (46). Comprehensive understanding of intermediary metabolic processes of intestinal microorganisms, and related isotopic fractionation pathways and patterns, including other potential mechanisms for sequestering light isotopes, however, is lacking for most vertebrate groups.

One taxon with a much lower than expected e*diet-bioapatite given its body mass was the giant panda (Ailuropoda melanoleuca). The substantial deviation from the predicted e*diet-bioapatite value likely is because the giant panda retains a carnivore-like gastrointestinal tract reflective of its ancestry as a member of the Carnivora, in spite of its evolution of an exclusively herbivorous, bamboo-specialist diet. Indeed, because of its long midgut but simple stomach, the giant panda lacks a digestive strategy that allows cellulose breakdown by means of the extended periods of gastric fermentation that is typically found in highly methanogenic mammalian herbivores.

Furthermore, the giant panda’s gut microbiota diversity is low and of a phylogenetic composition markedly different from that of other herbivores (i.e., lacking cellulose-degrading phylotypes) which is attributed as being responsible for its meager digestive efficiency (47).

The house mouse was the smallest (~20 gr) mammal evaluated, and its e*diet-bioapatite was lower than expected for its body mass. It is noteworthy that the other rodent analyzed, the prairie vole (~50 gr), also deviates from the regression model but instead exceeds the expected value. A comparative study of the fermentative structures (i.e., caecum + colon in hindgut fermenters) shows that the absolute and relative size of these structures in voles each exceeds (by more than

2 and 3 times, respectively) that of other rodents that are even two orders of magnitude larger in absolute body size (46). In voles, this relatively large gut capacity greatly increases the space available for microbial food fermentation, and might account for their higher than expected e*diet- bioapatite value relative to body size. Furthermore, voles possess a superior digestive efficiency relative to other rodents, correlated with a greater dry matter digestibility and selective retention of fine particles (48). Finally, in addition to possessing considerably smaller fermentative vats, mice need relatively higher rates of food turnover given their relatively higher metabolic requirements, leaving less time for food fermentation and therefore CH4 production. The type of

19 food selectivity (e.g., strict herbivory vs. omnivory) might also play a role in controlling e*diet- bioapatite variation. Comparative studies of guinea pigs (herbivore) and rats (omnivore) show lower methane production in the latter, explained as due to a lesser dependence on microbial fermentation due to their generally omnivorous diets (49). A comparable explanation may be applicable to the third Glires species analyzed, the European rabbit (Lagomorpha), which like the vole has a higher than expected e*diet-bioapatite value relative to its body mass, but a smaller residual from the overall regression than that of the vole. However, because the calculation for the rabbit was based on only one specimen (5), more data are needed to rule out the possibility that the higher than expected e*diet-bioapatite had been overestimated.

Besides the mouse and giant panda, the other two mammals with substantially lower e*diet-bioapatite relative to body mass were the three-toed sloth and koala, which share similar body masses (~4 kg), low metabolic rates, and strict folivory. Because the quality of ingested food is thought to influence microbial gut population activity (46), the additional metabolic challenges imposed by strict folivory (e.g., high content of secondary toxic compounds, low fiber content) would be expected to play a role in the e*diet-bioapatite observed in those taxa. In addition, both the three-toed sloth and koala feed on a small number of plant species. Koalas feed preferentially on

Eucalyptus spp. (50), whereas three-toed sloths eat Cecropia and Ficus spp. and in general are highly selective in their food choice, avoiding leaves of other tree species that are only moderately abundant on their home range (51). This suggests that both would have a specialized and less diverse gut microbiome (see (52)), and might also explain why the survival of Bradypus

(but not Choloepus) in captivity is so low (53). Finally, both the three-toed sloth and koala possess low basal metabolic rates and body temperatures (lower in Bradypus), which can result in lower rates of CH4 production by reducing the fermentation activity of gastrointestinal

20 symbionts (46). In marked contrast to the three-toed sloth, the two toed-sloth Choloepus has a larger e*diet-bioapatite than expected for its body mass. Major physiological differences between these two species could account for the substantial observed differences. For instance, although both species have fluctuating body temperatures in phase with that of their environments,

Bradypus shows an even more variable and lower overall daily body temperature than Choloepus

(51,54). In addition, although the field-measured metabolic rates of both sloths are the lowest of all non-hibernating mammals (54), both basal and field-measured metabolic rates are significantly lower in Bradypus (by one third) than in Choloepus (54,55). Moreover, unlike the three-toed sloth, which in the zoo was fed with a monospecific diet (sprouts of the rubber plant,

Ficus elastica), the food fed to the two-toed sloth included spinach and other greens, quinoa, sweet potatoes, carrots, and pellets of Purina Dog Chow® (table 1, ESM). In the wild, the diet of

Choloepus also is more diverse than that of Bradypus, even including components of animal origin (56–58). Consistent with a more diverse diet, Choloepus also has a more variable and diverse gut microbiota than Bradypus (52). Both a more diverse microbiome and a diet richer in fiber and fermentable carbohydrates seem to increase fermentation rates (46,52,59) and would explain the much higher e*diet-bioapatite values in Choloepus than in Bradypus, both absolutely and for that expected relative to body size for the correlation observed across mammals. Finally, an experimental study of Choloepus showed an unexpectedly high level of methane production in this species (higher than in non-ruminant herbivores of similar body mass), explained as due to long particle mean retention times (7). If so, this suggests that the mean retention times in the other extant sloth, Bradypus, is shorter.

Most artiodactyls and perissodactyls herein had e*diet-bioapatite closely correlated with their body masses (60). The only extant ungulate with a considerably higher than expected e*diet-

21 bioapatite value relative to its body mass was the cow, Bos taurus. In general, roughage-eating ruminants have the most complex digestive tract of all mammalian herbivores, with a well- developed forestomach, small caecum, and large populations of microorganisms in the rumen

(61,62). This type of ruminant digestive system involves an increase in time and capacity of fermentation, resulting in greater amounts of methane produced than a similarly sized non- ruminant, and a consequent expectation of higher e*diet-bioapatite value. In fact, the dietary energy lost through extensive microbial methane production in ruminants is the highest reported for modern mammals (61). Although precise controlled feeding data are lacking for a number of ungulates in this study, at around 100 kg regression models for foregut versus hindgut fermenters intersect with e*diet-bioapatite values of ~13‰. For mammals above ~200 kg, our model predicts higher e*diet-bioapatite values for foregut (including ruminants) compared to hindgut fermenters.

The African elephant, Loxodonta africana, shows a slightly lower e*diet-bioapatite than expected for its body mass. As mixed-feeder hindgut fermenters, elephants have faster digestive throughputs than ruminants, which results in less CH4 production (L/day) per unit body mass

(49). Strikingly, although the giant ground sloth Mylodon darwinii has an estimated body mass

(~1600 kg) only half that of an elephant, it possesses an even higher e*diet-bioapatite value, implying substantially higher methanogenic production. Because modern sloths carry out foregut fermentation and phylogenetically bracket all other sloths, it is most parsimonious to infer that fossil sloths such as Mylodon also had a foregut fermenting digestive tract. This type of digestive strategy would explain the high e*diet-bioapatite of Mylodon, as now expected given the steeper slope of the regression model for foregut fermenters compared to hindgut fermenters. In fact, our regression model for foregut fermenters alone predicts a higher e*diet-bioapatite than expected for hindgut fermenters, even excluding Mylodon from the analysis (figure 1.2). Ground sloths, the

22 largest mammals that ever existed in South America would be, together with modern hippos, the largest known non-ruminant foregut fermenters. Experimentally controlled e*diet-bioapatite data for modern hippos are not available, but our analyses across a wide range of mammals predict that hippos will show high e*diet-bioapatite values, at least comparable to those obtained for Mylodon.

Future studies will test if our predictions hold true and in what degree, if any, the combination of low rates of metabolism combined with amphibious lifestyles, diets of different nutritional quality and fiber content influence the expected e*diet-bioapatite in non-ruminant foregut fermenters such as hippos and other fossil giant sloths beyond Mylodon.

The predicted higher e*diet-bioapatite in foregut fermenters compared to hindgut fermenters across all mammalian herbivores analyzed is consistent with the considerable differences of CH4 production observed in foregut versus hindgut fermenters (60). However, methanogenesis does not explain the body mass effect observed in e*diet-bioapatite since there is an isometric relationship between body mass and methane production (60). This new body mass effect discovery thus requires a different, as yet unknown, size-dependent mechanism to explain diet-bioapatite C- enrichment. Therefore, although methanogenesis plays a role in differences of e*diet-bioapatite between digestive system fermentation types, it cannot explain the size effect of the relationship within each group or across all mammalian herbivores.

Figure 1.3: Correlations between e*diet-bioapatite and different life history traits. (a) Basal metabolic rate [BMR]. (b) BMR, after adjusting for the body mass component of the measured BMR (BMR/Kleiber value). The body mass component was removed by adjusting BMR to a value corresponding to 3.42*BM-0.25 (i.e., the Kleiber value (41)), indicating the scaling of BMR relative to body mass for mammals (body mass in grams (63)). (c) average rectal temperature, and (d) breadth of range of rectal temperature. All values in natural logarithmic scale. Temperature in °C. Dotted black line is the regression model for all taxa included; the red and greenish lines are the regression models for hindgut and foregut fermenters, respectively. Numbers for taxa same as in figure 2.

1.4 Significance of implementing body mass as predictor of e*diet-bioapatite

Because the standard value of 14‰ of diet-bioapatite enrichment has been widely applied across all mammals since the pioneering study of Cerling and Harris (8), with some notable exceptions (see above), our new equations incorporating the body-mass dependency for e*diet-

13 bioapatite most likely will change many previous reconstructions of dietary d C of mammals, particularly at the extremes of body masses: small-bodied and megaherbivore mammals. Initial examples of the utility of these new constraints can be drawn from the literature, in which

13 previous ecological interpretations using a e*diet-bioapatite of 14‰ lead to unreasonable dietary d C values. For instance, the dwarf antelope Neotragus batesi is a small (2-3 kg) artiodactyl (64) restricted to dense forests of tropical Africa. Dental d13C data of N. batesi from the Ituri Forest showed an average signature of -25.6‰. The use of the 14‰ e*diet-bioapatite enrichment reconstructed the d13C of the plant diet for this species as -39.6‰, lower than any plant ever recorded in that forest (d13C of subcanopy plants at Ituri range between -31‰ to -36.5‰ (1)).

Incorporating our new body-mass dependent enrichment, N. batesi, a small foregut fermenter,

13 would show an average e*diet-bioapatite value of 10.86‰, which indicates a plant diet with d C values of -36‰, a d13C signature observed for several plants sampled in that forest (1).

Recognition of the body-mass dependence of carbon isotope enrichment in mammals is essential for accurate understanding of the relative d13C differences among members of an herbivore community, which are used to reconstruct feeding niches and environments, and to evaluate ecological and environmental changes through time. Thus, species with similar d13C values, that would be interpreted as overlapping in resource usage when using a single standard e*diet-bioapatite value, could reveal distinct niche partitioning if body mass is taken into account.

Similarly, imprecise or even misleading reconstructions of dietary changes through time can 25

13 arise if the d C of the plants consumed have been inferred with a single e*diet-bioapatite value for all herbivores. For instance, application of our proposed body-mass-corrected e*diet-bioapatite could refine determination of the precise time when hominins began incorporating C4 plants in their diets and the proportion that these represented in the evolution of hominin diets over time.

Controlled feeding experiments may reveal the same body mass-specific enrichment between diet and other tissues, and this differential isotopic enrichment also could prove valid in non- mammalian organisms once tested through the same experimental methods.

Another important impact of our study would be on mean annual precipitation (MAP) calculations founded on d13C values of the plant community of the area under study (65).

Because the estimated d13C of the paleofloral community is indirectly inferred and calculated from d13C of dental bioapatite in fossil mammalian herbivores (65,66), a body-mass correction of

13 the e*diet-bioapatite can result in different d C vegetational values and a subsequent different MAP estimation. Accounting for the body mass influence on enrichment can profoundly alter some reconstructions of plant communities and their implied MAP values.

As the value of e*diet-bioapatite is related to digestive physiology, it can reveal physiological aspects in species that were not possible to assess previously, such as the extent of microbial fermentation to break up the ingested food, diet quality, and type of gastric fermentation (foregut versus hindgut). For example, our analyses indicate that the giant ground sloth Mylodon was a highly methanogenic foregut fermenter. Mylodon, a narrow-muzzled sloth interpreted as a selective feeder based on snout anatomy (67), is now documented as the mammalian herbivore with largest known e*diet-bioapatite value; therefore, we would expect other Quaternary fossil sloths that are twice as large as Mylodon (e.g., Megatherium americanum, Eremotherium spp.) to show even larger e*diet-bioapatite values, especially if they were bulk feeders, as suggested for the giant

26 ground sloth Lestodon armatus (67). Sloths challenge previously proposed ideas of constraints imposed by large body sizes on foregut fermentation (68) showing an unparalleled combination of morphological variability and physiological traits likely responsible for their extremely wide ecological diversification in the Americas.

Chapter 2: Comparative Isotope Ecology of Rainforest Mammals:

Amazonia Alters Current Paradigms

Julia V. Tejadaa,b,c,1, John J. Flynnb, Pierre-Olivier Antoined, Victor Pachecoc, Rodolfo Salas-

Gismondib,c,e, Thure E. Cerlingf

aDepartment of Earth and Environmental Sciences, Lamont-Doherty Earth Observatory, Columbia

University, New York, NY, USA; bDivision of Paleontology, American Museum of Natural History, New

York, NY, USA; cMuseo de Historia Natural, Universidad Nacional Mayor de San Marcos, Lima, Peru; dInstitut des Sciences de l’Évolution, UMR 5554, Université de Montpellier, Montpellier Cedex 5, France; eBioGeoCiencias Lab, Facultad de Ciencias y Filosofía, Universidad Peruana Cayetano Heredia, Lima,

Peru fDepartment of Geology and Geophysics & Department of Biology, University of Utah, Salt Lake

City, UT 84112

Tejada et al. in review

Climate and vegetation are traits defining tropical rainforests (69), but confident characterization and quantification of those archetypal traits of modern rainforests becomes increasingly challenging when analyzing ancient ecosystems. Closed canopy rainforests have been proposed to occur in the area now occupied by Amazonia since at least the Eocene, some

50 million years ago, but their extent, and the influence of the active Cenozoic geologic history in South America (with the Andean uplift as major driver) on these forests, is equivocal due to the sparse plant fossil record in the tropics and the variable interpretations from sedimentology and paleobotanical-based modelling (69,70). Isotopic evidence (d13C) from mammalian

28 herbivores can be a reliable proxy for reconstructing ancient ecosystems because the isotopic signals of vegetation (which tracks environmental factors and plant physiology) are recorded in primary consumers and are passed along the trophic chain to higher-level consumers. Thus, isotopic analysis is also a way to quantify vegetational criteria, which otherwise are primarily qualitative. To date, the isotopic structure of extant equatorial African mammals (1,10,71,72) has been the only broadly sampled system for understanding tropical ecosystems. Although phylogenetically and size biased, this African system has become the de facto model to infer closed canopy rainforests from fossil mammal data on all continents. However, the absence of comprehensive isotopic data from other closed canopy mammalian communities impedes confident attribution of an isotopic space for this type of ecosystem across landmasses and over time. Furthermore, is there a unique mammalian isotope space for all closed canopy rainforest ecosystems? Do different phylogenetic, geologic, and biogeographic histories influence the ultimate isotopic structure of an ecosystem? To answer these questions, we characterized the d13C structure of a modern western Amazonian mammal community in Peru. This rich mammalian community, the most diverse on the planet, exhibits a phylogenetic structure distinct from its African counterparts, resulting from tens of millions of years of geographic isolation over most of the Cenozoic in addition to distinct and complex biotic interactions. By comparing the isotopic structure of these two mammalian communities, equatorial Africa and Western

Amazonia, this study aims to define the d13C space for closed canopy rainforests across continents to serve as a baseline for understanding changes in the Amazonian ecosystem through time. Moreover, evaluation of resource exploitation within and among mammalian clades across both ecosystems can provide a new perspective on the ecological complexities of tropical South

American mammals, and potential influence of the Pleistocene extinction on niche occupation and its community structure.

2.1 South American and African mammalian communities

The history of South American and African mammals is intimately linked to the fate of these two landmasses after the break-up of the long-lasting southern supercontinent Gondwana.

South America and Africa, once part of Gondwana, fully separated by the mid-Cretaceous

(~110-100 Ma, (73,74)). South America, although briefly connected with North America at the end of the Cretaceous and with Antarctica until the early Paleogene, ultimately became and remained an island continent for most part of the Cenozoic (73,75). The phylogenetic structure of modern South American mammals is the result of tens of millions of years of geographic isolation, with a few exceptional trans-Atlantic dispersal events from Africa during the late

Paleogene, major faunal exchanges with North America during the late Neogene, and a dramatic extinction event at the end Pleistocene, in which more than 80% of mammals above 40 kg became extinct in South America (76,77). Africa, on the other hand, remained connected to

Arabia after Gondwanan breakup and drifted slowly northeastward, culminating in a collision with Eurasia in the late Eocene (78). What followed were largely intertwined evolutionary histories between African and Eurasian mammals. Notable is the Miocene-Pliocene radiation of endemic groups of African mammals including various bats, primates (e.g., hominids), hyracoids, carnivorans, proboscideans, etc. (79). In contrast to the dramatic events in South

America and most other continents, Africa was the continent least affected by the Pleistocene extinction event, with only 8 genera of megamammals disappearing (76).

2.1 Sampling localities and d13C of dietary sources of Amazonian mammals

The median d13C of terrestrial and aquatic C3 plants from western Amazonia fall within a range of -36.9‰ to -24.1‰ (80–83). Of these, leaves show the most negative signatures (-

32.1‰), compared to bole (-28.4‰) or litter (-28.7‰)(80). Although the data are not exhaustive, fruits and seeds (-29.3‰) show more positive d13C values than leaves (84). In a vertical profile of the forest, Amazonian plants show a gradient of decrease in their leaf d13C values with proximity to the ground, known as the ‘canopy effect’ (85). Indeed, leaf samples of plant species have average d13C values of -35.2‰ if growing within 1 meter above the ground, -33.4‰ in the lower canopy (2-10 m), and -30.5‰ in the upper canopy (>20 m) (82). These plant d13C values and the leaf d13C gradient are the same as those observed in African and other rainforests (1,86), as expected given the similar climatic and vegetational criteria that define all tropical rainforests

(69). Grass species are present in the Amazon rainforest and most of them utilize C4 photosynthesis (87). All but two specimens in this study are from localities in Peruvian western

Amazonia (mainly from the Madre de Dios, Ucayali and Loreto regions). Although some of these sampling areas (e.g., Manu National Park in Madre de Dios) span a large altitudinal gradient (Andean highland, cloud forest and lowland jungle), our sampling has been restricted to localities below 700 meters above sea level (i.e., lowland jungle, with the exception of 4 samples coming from the cloud forest). All samples, therefore, come from wet forest localities exhibiting analogous and homogeneous environmental conditions. Aiming to include all available data from the literature for other western Amazonian sites, two specimens of Tapirus terrestris from

Colombia and Bolivia have also been included (88). The African mammals assessed in this study come exclusively from equatorial lowland rainforests (primarily from the Democratic Republic of Congo, but also from Uganda and Gabon). We have not included data from any other African

31 ecosystem (e.g., savannas or woodlands). Our Amazonian localities in Peru span a latitudinal gradient of ~13° (from 0°40’S [Rio Curaray, the northernmost locality] to 13°6’S [Manu, the southernmost locality], although 68% of the samples come from Madre de Dios and Ucayali which encompass a 4° latitudinal range). Virtually all the samples come from undisturbed habitats in National Parks. The African localities span <5° of latitude (from 0.5°N [Kibale,

Uganda] to 3.3°S [Mwenga, Congo]).

Other dietary sources for some Amazonian mammals include aquatic plants, fishes, insects, or non-mammalian vertebrates. Aquatic systems in Amazonia are dominated by C4 macrophytes (mean d13C= -13.1‰, primarily represented by aquatic grasses); however, western

Amazonian fishes show a marked predilection for consuming C3 plants, with d13C signatures ranging from -37 to -21‰ (81,89). Herbivorous insects in Central Amazonia also exhibit a broad range of d13C variation (-29.5‰ to -15‰), generally mirroring their consumption of C3 or C4 plants (84,90). Bird data from Amazonian lowland rainforest indicate preferential consumption of C3 plants, whereas frogs and lizards derive at least half of their carbon from C4 sources

(90,91).

2.2 Materials and Methods

All but four specimens sampled are from closed-canopy rainforest habitats in western

Amazonian localities in Peru, including the well-known biodiversity hotspots Tambopata

National Reserve and Manu National Park, both in the Madre de Dios region. Specimens (only adults) were sampled from the mammalogy collections of the Museo de Historia Natural in

Lima, Peru, and the American Museum of Natural History in New York, USA. With few exceptions, only late erupting molars, ever-growing teeth, or canines were sampled. Our criteria for species selection are described in the SI Appendix. All samples were analyzed at the Stable

Isotope Research Facility at the University of Utah. We analyzed d13C from dental bioapatite

(enamel for all mammals, except xenarthrans [see below]) of 45 mammalian species (n= 176

13 15 13 13 individuals), d C and d N of hair keratin of 35 species (n=125), as well as d Cbioapatite, d Chair

15 and d Nhair from matched samples of a smaller taxon subset (31 species, n= 82). For the two

13 extant sloth genera, the d Cbioapatite was sampled from the outer dentine, whereas for anteaters and armadillos (toothless or with such small teeth that sampling was not possible) the d13C of the

‘enamel’ was projected from bone d13C values. We transformed bone d13C data to corresponding dental enamel values, using regression equations obtained from a separate analysis of the matched samples. All raw d13C data (for both bioapatite and hair) were corrected for

13 anthropogenic CO2 and set to preindustrial values (1750; d C1750). In order to make data

13 13 comparable among taxa, the CO2-corrected bioapatite d C was converted to dietary d C

13 (d Cdiet). For herbivores, we did so by using the body-mass dependent equations determined by

13 (92) to calculate the diet-bioapatite d C enrichment (e*diet-bioapatite) specific to each species (the calculated e*diet-bioapatite values for the herbivores in our study range from 10.3 to 13.7‰). For

13 secondary consumers the d Cdiet reconstruction used an e*predator-prey value of: -1.3‰ for carnivores (18), 7.8‰ for piscivores (93), and a theoretical 8‰ for insectivores. For herbivores,

13 d Cdiet refers to the calculated values for the vegetation on which they feed. For secondary

13 13 consumers (i.e., carnivores, piscivores, and insectivores), the d Cdiet refers to the predicted d C

15 of their prey. d Nhair data are presented as raw values (Fig. 2.3). All raw data and equations are presented in Table 2.1 and Dataset SI. Data from African mammals (10,71) also were standardized following the criteria described above. Mammals were classified into six dietary categories: herbivores (including browsers/grazers/folivores), frugivores, omnivores, carnivores,

33 piscivores, and insectivores. Except for omnivores, animals were binned into one of these categories when a dietary component (e.g., fruits for frugivores) represented >50% of the total diet. Statistical analyses, within and between orders, across continents, and between dietary guilds were conducted with both parametric (t-test, ANOVA) and non-parametric (Mann-

Whitney, Kruskal-Wallis) tests for significance. Unless noted, we report differences as statistically significant when p-values of pairwise comparisons for both parametric and non- parametric tests are £0.01.

RESULTS

13 2.3 d Cdiet of western Amazonian mammals

The taxa examined in this study included representatives of all non-volant mammal groups present in western Amazonia (terrestrial: Artiodactyla, Carnivora, Didelphimorphia,

Lagomorpha, Perissodactyla, Primates, Rodentia, Xenarthra [Pilosa, and Cingulata]; and aquatic:

Cetacea and Sirenia). This sampling yields the most complete d13C isotopic characterization of a closed canopy rainforest mammalian community to date. Our results document that western

13 Amazonian mammalian herbivores have a median d Cdiet of -27.4‰, ranging from -30.2 to -

12.3‰. This ~18‰ range of variation is bracketed by the white-tailed titi monkey Callicebus spp. and the tapir Tapirus terrestris at the lower end, and by the capybara Hydrochoerus at the upper end, the latter being the only C4 consumer of all the mammals analyzed (Fig. 2.1).

Figure 2.1: d13Cdiet (calculated from dental bioapatite) for all western Amazonian mammals analyzed. Boxplots represent the distribution of the data as displayed on the upper left corner. The d13Cdiet of the herbivores represent the vegetation on which they feed. Taxa inside red boxes are secondary consumers, so the d13Cdiet represents the predicted d13C1750 of the prey on which they feed. Numbers below the boxplots represent the number of samples analyzed per taxon.

13 Amazonian artiodactyls (4 species, n=16; deer and peccaries, median d Cdiet= -24.9‰), show an isotope span of 3‰, bracketed by the red brocket deer Mazama americana and the collared peccary Pecari tajacu at the lower and upper end respectively. Individuals of the

Amazonian tapir Tapirus terrestris, the only perissodactyl occurring in the study area, show little

13 isotope variation (<3‰), with a median d Cdiet of -28.6‰ (n=7). In contrast, the only lagomorph in this study (the tapeti, Sylvilagus brasiliensis) shows a large isotopic variation (5‰, median

13 d Cdiet= -28.2‰), even though the individuals sampled come from the same area (n=3). Rodents

13 (9 species, n=38) show d Cdiet values ranging from -28.8 to -12.3‰. This represents the largest

13 range of d Cdiet span (16.5‰) of all western Amazonian herbivore clades, driven by the

13 presence of the sole C4 consumer, the capybara (median d Cdiet= -15.5‰). Primates, the group best represented in our sampling (13 species, n=54), shows an isotopic span of 4.6‰, with

13 d Cdiet values ranging from -30.2‰ (Callicebus discolor) to -25.6‰ (black-headed night

13 monkey, Aotus nigriceps). Herbivorous xenarthrans (i.e., sloths) occupy a d Cdiet space of 3.6‰

13 and have a median d Cdiet= -27.8‰ (n=15). The only fully aquatic herbivore in our study, the

13 Amazonian manatee (Sirenia: Trichechus inunguis), has a median d Cdiet falling within the average values of terrestrial Amazonian C3 consumers (-26.6‰), and has a narrow intraspecific

13 d Cdiet variation (2.3‰, n=5). The only member of Carnivora that actually is a primary

13 consumer, the frugivore Potos flavus (kinkajou, n=5), has a median d Cdiet of -27.4‰ which differs significantly from that of the other terrestrial carnivoran species.

13 13 Reported d Cdiet values for secondary consumers reflects the d C of the prey on which

13 they feed (e.g., in the case of a carnivore feeding on an herbivore, it represents the raw d C1750 of the herbivore). For the clade Carnivora (excluding Potos), we applied an e*predator-prey= -1.3‰

(18), and an e*predator-prey= 7.8‰ for the aquatic predators in Cetacea (93). These secondary

13 consumers span an even larger range of d Cdiet variation than herbivores (22‰, from -32.9 to -

13 10.9‰, Fig. 1, Table 1). The lowest d Cdiet values are observed among the piscivorous taxa: the

13 giant otter (Carnivora: Pteronura brasiliensis; median d Cdiet= -27.8‰, n=6) and the Amazon

13 River dolphin (Inia geoffrensis; median d Cdiet= -31.9‰, n=2). This predicted dietary value for the piscivores is consistent with the very negative isotopic values of Amazonian fishes (81,89).

Terrestrial carnivores (Puma, Galictis, Atelocynus, and Leopardus) cluster at the upper end of the

13 isotopic spectrum with very high d Cdiet values (median= -13.7‰, SD=0.9‰, n=13). This

13 13 median d Cdiet (-13.7‰) for terrestrial carnivores matches the raw d C1750 values of artiodactyls and some primates, rodents, and lagomorphs in our study.

Dietary d13C reconstructions for the 7 insectivore species (one didelphid marsupial and six xenarthrans) was more challenging because no experimental e*predator-prey data are available for this feeding guild. We thus applied a theoretical e*diet-apatite of 8‰ for the insectivore-insect

13 13 d C offset (SI Appendix). The didelphid Philander sp. shows a median d Cdiet of -23.2‰ (n=4) and low variation among specimens (1.5‰). Armadillos and anteaters, the insectivorous

13 xenarthrans, have a dietary d Cdiet span of 2.5‰, ranging from -24.3 to -21.8‰ (n=8).

Acknowledging that future empirical experimental studies might document different e*predator-prey values for reconstructing insectivore diets, our related conclusions are conservative and do not

13 include the insectivore d Cdiet data.

Table 2.1: Summary of d13C results (‰) from dental bioapatite of western Amazonian mammals (A) and equatorial Africa (B) per taxonomic group. Mean d13C1750 refers to raw values corrected for anthropogenic CO2 and set to preindustrial values (1750) ± 1 SD. d13Cdiet refers to the reconstructed dietary d13C. For primary consumers, d13Cdiet refers to the vegetation on which they feed. For secondary consumers, d13Cdiet refers to the d13C of their prey. ‘Non- herbivores’ includes carnivores, piscivores, and insectivores. ‘Herbivores’ includes the other trophic group categories. C3= taxa consuming C3 plants; Herb Xen= herbivorous xenarthrans, Non-her Xen= non-herbivorous xenarthrans, Terr carn-carn= terrestrial carnivorous carnivores (i.e., excluding Pteronura [semiaquatic, piscivore] and Potos [frugivore]). Shaded cells: secondary consumers.

A. Western Amazonia

13 13 d C1750 Reconstructed d Cdiet #species/ Group specimens Median 13 13 d C1750 Median d Cdiet ± Range Range Span ± 1 SD span 1 SD All mammals 46 / 176 -15.9 ± -25.4 to 0.3 25.7 -27.2± -32.9 to -10.9 22 3.1 4.4 Herbivores 34 / 143 -16 ± -19.3 to 0.3 19.6 -27.4± -30.2 to -12.3 17.9 3.1 2.8 Non-herbivores 12 / 33 -15.4± 3 -25.4 to -12.2 13.2 -22.7± -32.9 to -10.9 22 6.4

Only C3 33 / 137 -16 ± -19.3 to -11.4 7.9 -27.5± -30.2 to -23.3 6.8 herbivores 1.7 1.5 Artiodactyla 4 / 16 -13.1 ± -15 to -12.2 2.8 -24.9± -27 to -24 3 0.7 0.8 Primates 15 /54 -16.8 ± -19.3 to -14.7 4.6 -28.2± -30.2 to -25.6 4.6 0.9 0.9 Rodentia 9 / 38 -15 ± -17.6 to 0.3 17.9 -26.1± -28.8 to -12.3 16.5 4.9 4.4 Lagomorpha 1 / 3 -17.3 ± -17.7 to -12.7 5 -28.2± -28.6 to -23.6 5 2.8 2.7 Perissodactyla 1 / 3 -14.9 ± -15.7 to -14.6 1.1 -27.9± -28.6 to -27.6 1 0.6 0.6 Sirenia 1 / 5 -13.2 ± -13.8 to -11.4 2.4 -26.6± -27.1 to -24.8 2.3 1.2 1.1 Xenarthra 8 / 23 -15.7 ± -17.2 to -14 3.2 -27.4± -29.4 to -21.8 7.6 0.8 2.6 Herb Xen 3 / 15 -15.9 ± -17.2 to -15.1 2.1 -27.9± -29.4 to -25.8 3.6 0.6 1.1 Non-herb Xen 5 / 8 -15.2 ± -16.3 to -14 2.3 -23± 0.8 -24.1 to -21.8 2.3 0.8 Carnivora 6 / 24 -15.9 ± -21.1 to -12.2 8.9 -14.8± -28.7 to -10.9 17.8 2.4 7.1 Terr carn-carn 4 / 13 -15 ± -16.4 to -12.2 4.2 -13.7± -15.1 to -10.9 4.2 1.1 1.1 Cetacea 1 / 2 -24.4 ± -25.4 to -23.3 2.1 -31.9± -32.9 to -30.9 2 1.5 1.4 Didelphimorphia 1 / 4 -15.4 ± -15.9 to -14.4 1.5 -23.2± -23.7 to -22.2 1.5 0.6 0.6 B. Equatorial Africa

13 13 d C1750 Reconstructed d Cdiet #species/ Group specimens 13 13 Median d C1750 Median d Cdiet Range Range ± 1 SD Span ± 1 SD span All mammals 31/140 -14.1± -24.5 to -0.9 23.6 -26.9± 4 -35.1 to -13.7 21.4 3.7 Herbivores 30/ 135 -14.1± -24.5 to -0.9 23.6 -26.8± -35.1 to -13.7 21.4 3.7 3.6 Only C3 28/ 124 -14.3± -24.5 to -9.8 14.7 -27± 2.3 -35.1 to -21.3 13.8 herbivores 2.3 Artiodactyla 20/ 87 -13.9± - 24.5 to -0.9 23.6 -25.8± -35.1 to -13.7 21.4 4.6 4.3 Primates 7/ 18 -15± 0.7 -16.1 to -13.2 2.9 -27± 0.6 -28.1 to -25.9 2.2 Proboscidea 2/ 32 -13.7± -16.6 to -11.1 5.5 -28± 1.2 -30.9 to -25.4 5.5 1.2 Rodentia 1/ 1 -16.3 - - -27.5 - - Carnivora 1/ 2 -11.7± -12.5 to -10.9 1.6 -10.4± -11.2 to -9.6 1.6 1.2 1.2

15 13 2.4 d Nhair and hair-bioapatite d C enrichment

Sampling of a proteinaceous tissue (hair keratin) in addition to bioapatite enabled us to

15 13 analyze natural variations in the d Nhair as well as a more nuanced view of dietary d C values among Amazonian mammals (Fig. 2.3, Table 2.2). Indeed, these data enable us to test whether the ‘expected’ offset between the d13C of bioapatite and that of a proteinaceous tissue (keratin in this study, e*bioap-keratin), proposed as indicative of trophic level (17,18), is met in this mammalian

15 assemblage. d Nhair values for 37 mammalian species span 10‰, ranging from Rodentia

15 (Dinomys branickii, median d Nhair= 2.3‰) to Carnivora (Pteronura brasiliensis, median

15 15 d Nhair= 11.8‰). The d Nhair values of the primary consumers are significantly lower than those of secondary consumers. There are no significant differences between bioapatite and keratin d13C enrichment (e*bioap-keratin) when binning taxa into primary and secondary consumers (Fig. 2.3A).

Herbivores (sensu stricto), however, do have significantly higher e*bioap-keratin values than frugivores, carnivores, and omnivores (at the 0.05 level), but do not differ significantly from insectivores or piscivores. Differences between other dietary group comparisons were not found.

13 Primary consumers drive the large variation in the range of d C e*bioap-keratin values across this mammalian community (>7‰, from 4.8 to 12.1‰, Fig. 2.3A). In contrast, secondary consumers cluster within a narrow range of e*bioap-keratin values (2‰, from 6.4 to 8.3‰). Frugivores show the

13 lowest e*bioap-keratin values of the entire mammalian community (range= 4.8 to 8.2‰). When d C enrichment between diet and keratin (e*diet-keratin) is assessed instead of e*bioap-keratin, we observe the expected clustering of carnivores at the lower extreme of e*diet-keratin values, and primary consumers with significantly higher e*diet-keratin values than carnivores. The e*diet-keratin among

39 primary consumers spans 6‰, ranging from 0.2 to 6.3‰ (Fig. 2.3B). Significant differences were only found in the e*diet-keratin between herbivores and frugivores.

Table 2.2: Summary of results for d15Nhair and d13Chair (standardized to 1750 values) and d13C e*diet-keratin for modern western Amazonian mammals. The reported range of values refers to medians per species (not specimens). Secondary consumers refer to carnivores, piscivores, and insectivores. ‘Herbivores’ include browsers, grazers, and folivores.

Group # median Range of median Median Range of 15 15 13 species d Nhair d Nhair d Chair 1750 e*diet-keratin e*diet-keratin All mammals 37 7.5 1.8 to 11.8 -23.3 3.3 -8.2 to 6.3 Herbivores 8 6.2 1.8 to 8.1 -23.6 2.4 0.2 to 5.1 Frugivores 14 5.4 3.8 to 10.1 -23.3 4.8 2.9 to 6.3 Omnivores 3 7.6 6.6 to 7.8 -23.6 4.8 3.3 to 5 Secondary consumers 12 9.5 7.4 to 11.8 -22.4 0 -8.2 to 1.6 Artiodactyla 1 1.8 - -24.4 4.7 - Primates 14 5 3.8 to 8.6 -23.5 4.7 2.9 to 6 Rodentia 8 7.6 2.3 to 10.1 -22.9 2.2 0.2 to 6.3 Lagomorpha 1 3.1 - -25.3 2.9 - Xenarthra 7 7.9 7 to 9.5 -22.8 1.6 -0.3 to 5.1 Carnivora 5 10.4 5.4 to 11.8 -21.8 -7.7 -8.2 to 4.9 Didelphimorphia 1 9.5 - -23 0.2 -

DISCUSSION

13 2.5 Comparisons of d Cdiet of mammalian herbivore communities from western

Amazonia and equatorial Africa

Only one C4 consumer was identified in western Amazonia whereas at least three C4 specialists exist in African rainforests (Fig. 2.2, SI Appendix Fig. S3). None of the Amazonian species analyzed (representing >90% of all herbivores above 1 kg body mass in the study area) fills the carbon isotopic niche occupied by African understory forest dwellers. The breadth of

13 isotope d Cdiet values exhibited by herbivorous mammals in both continents is comparable (21‰ vs 18‰ in Africa and SA respectively), but while the isotopic range in Amazonia is driven primarily by the positive signatures of the only C4 consumer (the capybara), in equatorial Africa

13 this similarly broad range is driven instead by the extremely negative d Cdiet values (<-30‰)

40 observed in the few species feeding in the subcanopy stratum of the forest (the antelope

Neotragus batesi [Bates’s pygmy antelope], the giraffid Okapia johnstoni [okapi], and some individuals of the suid Hylochoerus meinertzhageni [giant forest hog] among artiodactyls, and the forest elephant Loxodonta cyclotis). In Amazonia, no terrestrial mammal exhibits such

13 extreme negative d Cdiet values, even though species living and feeding in the subcanopy stratum are represented in our analysis, and plants with d13C values as negative as the most negative plants in Africa do exist in western Amazonia (80,81). Within a forest, the most negative d13C values are found in leaves growing in the understory, where light deprived conditions increase isotope discrimination (94,95). In fact, a d13C difference up to almost 5‰ can be seen within a single plant species, and a range of 10‰ can occur across a vertical profile of the canopy, depending on the amount of light received (86,94). Therefore, for mammals to record such negative isotopic values, they have to be selective in which plants they eat, but most

13 importantly, where they do so. This foraging selectivity limits this extremely negative d Cdiet niche to selective-feeding forest dwelling herbivores, i.e., mammals almost exclusively eating

13 leaves growing deep in the understory. That mammals with median d Cdiet values <-30‰ are rare even in Africa, the only modern ecosystem where these values have so far been identified, indicates that these signatures cannot be extrapolated as the “expected” isotopic signatures for all mammals living in the subcanopy stratum of any rainforest nor to be used as an indispensable indicator of rainforests. Subcanopy-feeding herbivorous mammals in western Amazonia today, therefore, are consuming vegetation falling from the upper layers of the canopy, or consuming items with a wide range of isotopic signatures (e.g., leaves growing under different degrees of canopy closure, or incorporating a variety of food items in their diets), thereby averaging the d13C of their energy pool to yield values closer to the typical median for rainforests.

Figure 2.2: d13Cdiet values and distributions of the herbivore mammalian communities in western Amazonia (left) and equatorial Africa (right). Histograms on left axis represent the d13C values for plants with C3 and C4 photosynthesis (modified from (96)). Numbers below boxplots represent the number of samples per taxon.

13 The median d Cdiet of artiodactyls in South America and Africa are not significantly different (Table 2.1). The isotope range exhibited by artiodactyls in Africa (>21‰, ranging from

-35.1‰ to -13.7‰), however, is significantly larger than in South America (5‰, ranging from -

29‰ to -24‰). This dramatic difference may result from the disparity in diversity of the artiodactyl clade samples across the two continents (19 vs 4 species in our dataset, in South

America and Africa respectively), but we note that artiodactyls also are much more diverse in

Africa due to their long history on that continent and the extremely recent arrival of the group in

13 South America. The d Cdiet of primates in both continents differs significantly, but the

13 difference in means is only 1‰ (mean d Cdiet= -28‰ for SA primates, -27‰ for African primates) and fewer species were sampled in Africa, so this difference could be reduced or

42 disappear with a more extensive sampling of the many other African rainforest primate species.

Rodents are poorly sampled in equatorial Africa (data were available for only one species, the phiomorph Atherurus africanus [African brush-tailed porcupine]), but in western Amazonia

13 13 rodents span a broad d Cdiet range (16.5‰), mainly driven by the d Cdiet of the capybara, whose

13 13 exclusion decreases the breadth of d Cdiet variation in rodents to 5.5‰. This wide d Cdiet niche occupation might also reflect the ecological diversity resulting from a relatively long evolutionary history of caviomorphs on the continent (see last discussion section). With this rationale, we also might expect Xenarthra (along with the marsupials, all of which are secondary

13 consumers) to occupy a large range of d Cdiet values because these are the only other survivor groups from the original pool of mammals that existed in South America before the faunal immigration waves that occurred throughout the mid-late Cenozoic. Modern herbivorous xenarthrans (i.e., sloths), however, represent only 2% of the earlier diversity of the group in the

13 fossil record, and our results document that modern sloths are restricted to a very narrow d Cdiet range (<4‰), consistent with their low modern taxonomic and ecological diversity (limited to three species in western Amazonia).

13 Reconstructed d Cdiet for terrestrial carnivores (median= -13.7‰, excluding the

13 frugivore Potos and the semiaquatic Pteronura) match the raw d C1750 values of artiodactyls, as well as of some primates, rodents, and the only lagomorph species in our study. This

13 reconstructed d Cdiet relies, however, on a e*predator-prey= -1.3‰ inferred from only specialized hypercarnivores (18). This e*predator-prey= -1.3‰ might be unrealistically low, because Amazonian mammalian predators are rather opportunistic in their feeding behaviors (including insects, other non-mammalian vertebrates, and plant elements in their diets). A more positive e*predator-prey value

13 13 would make the d Cdiet of these faunivores match the d C1750 of most herbivores in this study,

43 as it also could for the African genet Genetta, for which a e*predator-prey= -1.3‰ may not reflect the omnivorous feeding behavior of this species.

The three aquatic mammals (Cetacea: Inia geoffrensis, Carnivora: Pteronura brasiliensis,

13 and Sirenia: Trichechus inunguis) exhibit d Cdiet values that all correspond to exclusive consumption of food with carbon sources of C3 origin (Fig. 2.1). Among these, the two

13 piscivores (river dolphin and giant otter) show significantly different d Cdiet values (medians of -

32‰ and -28‰ respectively), with that of Inia being the most negative of all the species

13 sampled. The difference in d Cdiet values of these two species might be due to Inia being an exclusive piscivore, whereas Pteronura is not. Indeed, Pteronura incorporates other vertebrates and invertebrates in its diet (97), both of which have more positive d13C values than usually observed in Amazonian fishes. This broader feeding choice in Pteronura also is consistent with a

13 larger intraspecific d Cdiet variation (Fig. 2.1, Table 2.1).

One key question in our study was whether or not it is possible to define a closed canopy

13 rainforest from mammalian isotope data. Our results show that the median d Cdiet of terrestrial herbivores in South America (-27.4‰) and Africa (-26.9‰) are not significantly different. The

13 median d Cdiet for these two rainforests is -27.2‰ (SD= 3.2‰, SE= 0.2‰), a value that could be expected for mammalian herbivores living in any closed canopy tropical rainforest. Although

13 data are limited, the mean d Cdiet of non-rainforest mammalian communities seems to be higher than -27.2‰ (SI Appendix, Figs. S8 and S9). The -27.2‰ value for closed canopy tropical rainforests is nearly identical to the global mean d13C for plants (98). Our results show, however,

13 13 that the d Cdiet from mammalian data should not be expected to exactly reflect the d Cplant

13 values, as the mean d Cplants of tropical rainforests is more negative than the global mean (-

13 31‰). That both Amazonian and African herbivores show an offset in their d Cdiet relative to

13 the overall rainforest Cplant values (medians -27.2‰ vs -31‰ respectively), documents complexities in the incorporation of carbon from diet to tissues and the necessity of comprehensive baseline studies to understand the processes underlying this offset and better characterize the isotopic structure of these ecosystems.

2.6 d13C enrichment between proteinaceous tissues and bioapatite

Increases in position within the trophic chain have been linked to stepwise rises in d15N values, with 3-4‰ as the constant value usually invoked for each change in trophic level (19,99).

However, substantial variation in d15N values among trophic guilds, and unexpectedly high d15N values in some herbivorous species (overlapping that of carnivores), also have been identified

(18,19), making the use of d15N alone a potentially imprecise or even misleading proxy to identify trophic levels (without knowledge of d15N baseline values and especially if comparing taxa among habitats). Instead, case studies document that enrichment between the d13C of bioapatite and a proteinaceous tissue (e*bioap-prot) is a more reliable method for assessing food chain relationships (17,18). Carnivores and herbivores are expected to have opposite values within the spectrum of e*bioap-prot (low and high, respectively), while omnivores show intermediate values. The rationale behind this expectation is that the type of digestive system influences the degree to which food is degraded by fermentation or by endogenous enzymes

(affecting the subsequent degree of d13C transformation of bioapatite relative to diet), and that the contribution of different macronutrients (carbohydrates, proteins, and lipids) in the synthesis of bioapatite and proteinaceous tissues differs according to feeding behavior and food choice

(17,100,101).

Our results show that this expectation is simplistic in a hyper diverse natural environment like Amazonia, revealing intricacies associated with the breakdown of dietary macromolecules in

45 a community of primary consumers mostly characterized by generalist taxa (i.e., ‘herbivores’ that incorporate a wide variety of plants, and some even animal elements, in their diets). Indeed, of the 36 species of primary consumers analyzed, only six can be classified as obligate herbivores (i.e., animals whose fundamental niche precludes them from incorporating any animal tissues in their diets: Bradypus, Trichechus, Sylvilagus, Hydrochoerus, and two species of

Coendou). Of these, only Bradypus and Coendou are classifiable as obligatory specialists (i.e.,

13 very narrow realized niche and diet (102)), corroborated by their narrow intraspecific d Cdiet values. Contrary to findings in other studies (e.g., (18)), frugivores rather than secondary consumers had the lowest e*bioap-keratin values, a result that could be explained by the high lipid content of Amazonian fruits and seeds (103). Body protein tracks dietary protein, whereas bioapatite tracks bulk diet (i.e., the combination of all three macronutrients: carbohydrates, proteins, and lipids). Diets with high lipid content will decrease the d13C of the whole diet,

13 13 because lipids are C-depleted relative to other macronutrients. This more negative d Cdiet will

13 13 then be imprinted in the d Cbioapatite (thus now closer in its d C value to proteinaceous tissues).

Even though low e*bioap-keratin among some frugivores might also be interpreted as reflecting omnivory, in the current paradigm, omnivores would not be expected to show lower e*bioap-keratin values than true carnivores or insectivores. Alternatively, these results can be explained with an utilization of macromolecules and energy ratios that are inconsistent with traditional herbivore macronutrient profiles, as recently identified in an obligate specialist herbivore (104). Indeed, that study (104) revealed that by switching foraging areas associated with asynchronous phenologies of two bamboo species, giant pandas maximized their protein intake and minimized their fiber ingesta. This resulted in a dietary macronutrient composition that equates to that of hypercarnivores because of similar reliance on proteins, albeit plant rather than animal, as the

46 dominant macronutrient source. With such a high percentage of energy coming from dietary protein (instead of carbohydrates, as in most herbivores), we also predict that pandas will show low values in the e*bioap-prot spectrum, when measured. The lower than expected values in e*bioap- keratin observed in some Amazonian herbivore species might be explained in a similar way. Other studies have also shown highly variable e*bioap-protein values within a free-ranging herbivore mammalian community (105), as well as e*bioap-protein poorly distinguishing trophic levels (106), suggesting that the e*bioap-protein within dietary categories (especially among primary consumers) should not be expected to be uniform.

Figure 2.3: (A) d15Nhair vs d13C e*bioapatite-keratin of Amazonian mammals. (B) d15Nhair vs d13C e*diet-keratin (d13Cdiet from bioapatite) for Amazonian primary consumers only. Each point represents the median value per species.

The lack of trophic level segregation in the e*bioap-prot spectrum leads to reconsideration of tacit assumptions underlying this expectation: (1) diet-bioapatite enrichment within the herbivore primary consumer guild is not significantly different among species, (2) the general dietary macronutrient profile of herbivores and carnivores is always different, and/or (3) d13C enrichment between animals’ proteinaceous tissues and dietary protein is relatively constant.

The invalidity of the first assumption becomes evident when plotting enrichment between diet and keratin (e*diet-keratin, Fig. 2.3B) rather than between bioapatite and keratin. Indeed, after correcting for diet-bioapatite enrichment values, we observe the expected segregation of carnivores at the lower extreme of e*diet-keratin values and primary consumers showing significantly higher e*diet-keratin values. Among primary consumers, the obligate herbivores

Hydrochoerus and the two Coendou species (Rodentia) show the lowest e*diet-keratin values (Fig.

2.3B). Even using a higher e*diet-bioapatite value (e.g., the older uniform value of 14‰) than our body mass-corrected values to reconstruct the dietary d13C (11.8‰ for 4 kg Coendou and 12.7‰ for 50 kg Hydrochoerus) does not place these species at the uppermost extreme of the e*diet-keratin spectrum, as would be expected for obligate herbivores. Furthermore, the specialized obligate folivore Bradypus, a species with a known controlled-feeding e*diet-bioapatite value (92) also shows e*diet-keratin values at the lower end of that for all primary consumers, suggesting that dietary traits

(in addition to or rather than physiological traits) might be involved instead.

That many Amazonian mammals do not fall in the ‘expected’ place in the e*bioapatite-keratin spectrum, when classified by their feeding choice, calls into question the predictive power of traditional dietary ecological classifications for bioapatite-protein isotopic offset expectations.

Analysis of the hyperdiverse Amazonian closed-canopy rainforest mammal community shows that the widely held assumption of distinct macronutrient profiles between carnivores-herbivores and specialists-generalists is not always met (104,107), and a better characterization of the mixture of nutrients in an organism’s diet (rather than just the kinds of food or energy content) are necessary to fully understand diet-tissue isotopic fractionations. Indeed, the large range in the e*diet-keratin among Amazonian primary consumers illustrates previously unexpected complexities associated with routing of macronutrients for protein synthesis. The low e*diet-bioapatite values in

48 obligate herbivores like Hydrochoerus, Coendou, Dinomys, or Bradypus suggest that dietary proteins are supplying their amino acid needs for keratin synthesis. In contrast, the large e*diet- bioapatite values observed in most frugivore species suggest that these species are synthesizing amino acids de novo (likely from carbohydrates) to produce keratin (Fig. 2.3B).

2.7 Potential influence of Pleistocene extinction on niche occupation of Amazonian mammals

“… If Neotropical ecologists and evolutionary biologists wish to determine who eats fruit, who carries sticky seeds, and who browses, grazes, tramples, and voids that segment of the habitat that would have been within the reach of a variety of megafaunal trunks, tusks, snouts, tongues, and teeth, the missing megafauna must be considered…” Janzen and Martin, 1982.

“Neotropical anachronisms: the fruits the gomphotheres ate”

Amazonia is the world’s largest rainforest, and western Amazonia in particular is further considered to harbor the highest modern mammalian diversity on the planet (108,109). Yet, the isotopic space occupied by mammalian herbivores there is narrower than that of equatorial

Africa, even though the sampled Amazonian localities span a wider latitudinal range and

13 Amazon closed-canopy rainforest vegetation exhibits a similar d Cdiet range to that observed for

African plants. Why do equatorial African mammals exploit a broader spectrum of resources than Western Amazonian mammals? Why do Amazonian mammals not consume all available plant resources in the forest, instead occupying a comparatively narrower breadth of isotopic

13 niches than in Africa? The d Cdiet data from terrestrial equatorial African mammals indicate that four artiodactyl species exploit resources at the isotopic extremes in a closed canopy rainforest, among these are two pure C4 consumers (Syncerus caffer nanus and Phacochoerus africanus)

13 and two forest dwellers with extremely negative d Cdiet values (Neotragus batesi and Okapia

49 johnstoni, although some individuals of the suid Hylochoerus meinertzhageni as well as the

13 forest elephant Loxodonta cyclotis also show d Cdiet< -30‰). In South America, only one rodent species occupies the upper isotopic extreme (C4 consumer space) and no Amazonian mammal seems to be feeding (at least exclusively) on the most isotopically negative plants (i.e., in the isotopic space occupied by Neotragus and Okapia in Africa). Assessing these differences requires comparison of the herbivorous mammalian communities in South America and Africa in biological traits that might influence isotopic niche occupation. One important distinction is substrate occupation. In Africa, 73% of the species sampled are obligately terrestrial, whereas these represent only 39% of the Amazonian sample. This is relevant because terrestrial mammals are those most likely to feed on plants growing in the lowest stratum of the forest, the understory.

The difference in the number of terrestrial Amazonian species is not a flaw in our sampling design; there simply are fewer exclusively terrestrial mammals in modern Amazonia than in

Africa, and particularly in the much lower number of ungulate (artiodactyls, perissodactyls) species. A similar situation pertains for numbers of obligate herbivores (~14% in Amazonia vs

13 >60% in equatorial Africa), which also might influence the smaller breadth of d Cdiet values observed in Amazonian mammals. Therefore, arguably the most important variable, one that encompasses the biological traits differing between the mammalian communities of these two tropical rainforests (e.g., substrate occupation, feeding niches, body mass), is the distinct evolutionary time, and therefore feeding guild scope, represented by clades in both continents

(Fig. 4). Indeed, while the evolutionary history of most lineages of modern terrestrial African mammals can be traced back to the Paleogene, only caviomorph rodents can be considered as terrestrial herbivores native to South America prior to the late Pliocene-Pleistocene Great

American Biotic Interchange (GABI). And caviomorphs represent, in fact, the group of

13 mammals with the largest d Cdiet range in Amazonia. Given that species diversification is often followed by niche evolution (110), and that diversification is related to evolutionary time (111), time may be an important factor in determining the breadth and variety of isotopic fundamental niches that species within a mammalian herbivore community can exhibit. The modern mammalian communities in equatorial South America and Africa therefore are not strictly comparable ecologically, because the former represents a post-extinction ecosystem, whereas the latter was not comparably affected. By eliminating all of the ancient Paleogene lineages of endemic South American terrestrial herbivorous mammals (e.g., terrestrial sloths, and now entirely extinct South American native ungulates [notoungulates, litopterns]), as well as many groups of large herbivores that arrived only during the GABI (e.g., proboscideans, perissodactyls, etc.), the Pleistocene extinction event dramatically reduced the evolutionary time

(phylogenetic scope) represented by survivors in the modern Amazonian mammalian community. Thus, even though our sampling of extant Amazonian herbivores (terrestrial and non-terrestrial) better encompasses the total phylogenetic diversity of the ecosystem than does the African sample, the shorter evolutionary time and restricted phylogenetic breadth represented by modern South American mammals arguably explains the more restricted isotope space occupation compared to Africa (Fig. 2.4).

Figure 2.4: d13Cdiet values for both western Amazonian and equatorial African mammals plotted into their respective phylogenies. Although the medians for d13Cdiet in western Amazonia (-27.4‰) and equatorial Africa (-26.9‰) are not significantly different (median for the two= -27.2‰, red box on the d13Cdiet axis), the d13Cdiet space occupied by mammalian clades differs dramatically in each rainforest.

Although it remains to be tested, we propose that sampling some of the mammals that became extinct in the Pleistocene might broaden the spectrum of isotopic niches observed in modern Amazonia (e.g., recovering some of the isotopic values <-30‰ observed in Africa, and reflecting that of known Neotropical plants). This range of ecologies that does not currently occur in Amazonia could have been occupied by clades of terrestrial mammalian herbivores with no close extant relatives (e.g., notoungulates, litopterns) that are known to have gone extinct recently, groups of herbivores with unparalleled physiological traits (e.g., extinct giant ground sloths, the largest foregut fermenters that have ever existed), and other mammals that occupied currently empty body mass categories (e.g., >200 kg). Analysis of the large-bodied Pleistocene herbivore Toxodon (a notoungulate, one of the 66 or more megafaunal species that became extinct in the Pleistocene of South America (112)) from a broad latitudinal range in the Americas showed exclusive consumption of C4 plants at high latitudes but C3 plants in the Amazon (113).

13 13 The calculated d Cdiet of Amazonian toxodonts (median d Cdiet= -27‰) was found to be lower

52 than modern Amazonian artiodactyls, although still not as negative as the African subcanopy

13 feeders (i.e., d Cdiet< -30‰). Although other studies have isotopically characterized Pleistocene taxa at high latitudes in South America (114,115), no isotopic characterization of Pleistocene mammalian communities from closed-canopy rainforests has yet been done because of limited known localities from these habitats, where plants with d13C <-30‰ are present and may have been consumed by recently extinct herbivore lineages, as is observed in equatorial Africa today.

Isotopic characterization of the Pleistocene Amazonian mammalian community also could better inform understanding of the influence of the extinct megafauna on the realized niches of modern

Amazonian mammals, by revealing if surviving lineages shifted or expanded their feeding ecologies (and isotope niche occupation) after the extirpation of those species from the ecosystem. Indeed, Pleistocene megafaunal extinction was suggested to be responsible for

13 expansion of modern deer into their current d Cdiet niche within temperate North American habitats (116). Alternative potential explanations for the restricted isotopic occupation observed in the modern Amazonian mammalian herbivore community might include inherent traits reflecting their narrower feeding diversity compared to equatorial African mammals, or a sampling bias. Given that isotopically similar plant resources are available in both rainforests

(C4 grasses, and plants with d13C<-30‰), that Amazonian mammals might specifically avoid consuming these resources would be intriguing. Our sampling is limited for mammals <0.3 kg, which might be consuming the highly negative understory plants, but includes more than 80% of all larger species in the sampling area, spanning the body sizes of taxa with highly negative values in Africa. Excluding marsupials, which are omnivores, the only terrestrial herbivorous mammals <0.3 kg are rodents, and small rodents are not represented at all in the African sample.

Future empirical geochemical sampling and paleontological field efforts in pre-Holocene

53 deposits of Amazonia will test our hypothesis of the impact of Pleistocene extinctions and alternative explanations, and should reveal new questions involving complexities of ecological interactions over time in what was perhaps the most biodiverse continental ecosystem in Earth history.

Chapter 3: TESTING DIETARY-CONSUMER d15N OF AMINO

ACIDS IN MODERN XENARTHRAN SPECIES AS A BASELINE

FOR INTERPRETATION OF FOSSIL SLOTH AMINO ACID

COMPOUND SPECIFIC DATA

Home to more than half of Earth’s land biota, South America contains the largest diversity of terrestrial mammals (109). Little is known, however, of how mammalian communities were ecologically structured before the megafaunal extinctions of the last ice age

(Pleistocene) where more than 80% of mammals above 40 kg became extinct (112). Indeed, the paleoecological characterization of South American (SA) fossil mammals is particularly challenging because their communities have no full analogs in modern or fossil guilds from other continents due to marked taxonomic distinctions and large numbers of entirely extinct clades

(117). Part of the explanation for this perceived uniqueness is the depauperate abundance and diversity of its mammalian carnivore fauna throughout the Cenozoic, as suggested by a number of analytical methods. Allometric functions relating population densities and body mass have suggested that pre-Pleistocene SA mammalian communities were ecologically imbalanced, because the biomass of herbivores significantly surpassed the energetic requirements of carnivores (22). Likewise, it has been suggested that fossil ecosystems could not have supported the plant biomass demands of all of the presumed large herbivores, and as a consequence there must be secondary consumers disguised as herbivores in the fossil record (22). Because of their simple (and potentially functionally versatile) dentitions and high taxonomic diversity, fossil sloths have been identified as the most likely candidates to have occupied scavenger niches (22).

Modern sloths, however, are obligate tropical arboreal herbivores, an extreme case of feeding specialization involving the exploitation of food of low nutritional quality, with the extra physiological challenges entailed by the poor digestibility of plant allelochemicals (92). The traditional consensus is that fossil sloths also were obligate herbivores like their modern relatives

(mostly based on craniodental morphology analyses and a presumption of common dietary adaptation across the group). Yet their extremely high diversity and incredibly broad range of niches, body sizes, and inferred habitats, including even a marine sloth (118), suggest that the hundreds of known species of fossil sloths were ecologically much more versatile than traditionally thought. Considering that xenarthrans (the group including modern sloths, anteaters, armadillos, and a many extinct close relatives) comprised an important proportion of South

American mammalian diversity until as late as 12,000 years ago (25), evaluating xenarthrans’ trophic relationships is crucial for understanding the evolution of Neotropical mammalian biodiversity as a whole, as well as the interplay between resource availability, ecological functions among xenarthrans, and competition within and across mammalian clades in the continent over the Cenozoic.

Isotopic analyses are now widely used in ecological and paleoecological studies. These techniques rely on two underlying premises. First, that isotopic values of any given chemical element differ across diverse pools and reservoirs in the biosphere, and therefore that different foods and waters are isotopically varied. Second, because an individual’s body tissues are synthesized from the food and water they ingest, their body tissues are isotopically linked in a predictable way to their food/water intake (119). Thus, stable isotope analyses offer a rare route to document the consumed diet itself, rather than through indirect inferences based on morphological indicators, representing a particularly useful and potentially more reliable method

56 to be applied to fossil samples. Traditionally, bulk nitrogen stable isotope analyses (d15N) have been used to evaluate trophic interactions due to an often-found increase of 3-5‰ in d15N per trophic level change (17,18). However, substantial variation in d15N values has also been observed among trophic guilds (18,19), indicating that the use of d15N alone can be a misleading proxy to identify trophic levels. Indeed, interpretation of bulk d15N often fail to acknowledge variations in baseline d15N values within an ecosystem, which can greatly differ depending on the way primary producers synthesize nitrogen and the isotopically different types of inorganic

- nitrogen sources available (e.g., N2, NH3, NO 3, (20)). Amino acid compound specific isotope analysis (AACSIA), a more recent technique that measures and evaluates the d15N of specific amino acids, overcomes the issue of unknown or variable baseline d15N values because it records both the baseline and the trophic d15N in the same organism (12,16,28,120,121). Indeed, studies have shown that certain amino acids experience little change in their d15N values from diet to consumer (12,120). In contrast, the d15N of some other amino acids differ significantly in consumers’ tissues relative to their diets. The explanation behind this observation relies on the extent to which amino acids exchange their amino-nitrogens with the overall metabolic pool of nitrogen, and subsequent minimal or large d15N isotopic fractionations experienced during metabolism (15). Amino acids that readily exchange their amino-nitrogens and experience large isotopic fractionations have been termed “trophic” amino acids (i.e., glutamic acid, aspartic acid, alanine, isoleucine, leucine, proline, valine), whereas those that experience minimal exchange of their amino-nitrogens, and are closer in their d15N values to that of the diet, are known as

“source” amino acids (15,16). The higher the trophic position of the organism under study, the greater the number of metabolic cycles in which their amino-nitrogens will be involved

(especially the ‘trophic’ AAs), and therefore, the greater the expected isotope fractionation.

Among ‘trophic’ AAs, glutamic acid (Glx) has been observed to experience a constant change in its d15N with each trophic level transfer relative to the original dietary Glx d15N value, which is why it is preferred over other “trophic” amino acids for calculating trophic positions. Likewise, among source amino acids, phenylalanine (Phe) has been observed to experience a constant minimal change in its d15N relative to dietary Phe values which can explain its use over other

“source” amino acids (28). Glutamic acid and phenylalanine have thus become the two canonical

“trophic” and “source” amino acids, and thus are used to calculate the trophic positions of organisms (12,14,16,28,120,121), although a few studies have instead proposed a combination of trophic and source amino acids to calculate trophic levels (20,122).

The application of AACSIA for ecological purposes, however, is not without caveats, as determination of food web dynamics using these methods relies on some poorly tested assumptions. The foundational studies of AACSIA were based mainly on marine ecosystems, and the validity of the method for application in terrestrial ecosystems has not yet been thoroughly evaluated. Furthermore, the technique relies on the assumptions that the isotopic differences among glutamic acid and phenylalanine in primary producers (β value), and the 15N- enrichment of glutamic acid and phenylalanine between consumer and diet (trophic discrimination or enrichment factor, TDF or TEF), both are constant across all organisms. A β value of -8.4‰ has been accepted as representative of C3 ecosystems, whereas the standard TDF value used to calculate trophic positions of all organisms is 7.6‰ (28). These values already are being used to calculate trophic positions of terrestrial vertebrates, including mammals, even though they have not yet been thoroughly examined in terrestrial ecosystems (indeed the β value of -8.4‰ is based on a few number of plant species, whereas the TDF value is based solely on marine invertebrates, fishes, and insects (12,121). The technique is been used in archaeological,

58 anthropological, and paleontological research (e.g., to evaluate dietary practices of ancient humans and other hominid species (e.g., (14)). Yet, the only controlled experimental study on a mammalian species, Bos taurus, (123), showed different β and TDF values than those used in the

Chikaraishi et al. (2011) equation based on marine systems, suggesting that these values cannot be applied directly across food webs and ecosystems.

The objectives of this chapter, therefore, are twofold. First, it aims to rigorously test a controversial hypothesis of carnivory and consumption of proteins of animal origin on two fossil sloth species by means of d15N AACSIA: Mylodon darwinii and Nothrotheriops shastensis. But in order to confidently achieve that goal using this analytical technique, I will first test if the underlying assumptions behind the method hold true for modern sloths and other closely related species under different, known feeding regimes. Thus, this project has two stages:

1. to test whether there is a consistent relationship between the d15N of dietary and

consumer amino acids for different modern xenarthran species under different feeding

regimes, to provide a potentially reliable baseline for interpretation of d15N AACSIA of

fossil samples, and

2. to test previous hypotheses about the feeding ecology of two fossil sloth species,

Mylodon darwinii and Nothrotheriops shastensis, through the AACSIA d15N analyses,

and determine whether meat might have been a significant component of the diet of these

presumed herbivore specialists.

MATERIALS AND METHODS

3.1 Animal and diet samples

Hair samples for two Pleistocene fossil sloth species (Mammalia: Xenarthra) were included in this study: Mylodon darwinii (Mylodon Cave, Chile, 2 specimens, 3 samples, AMNH

FM 96263, AMNH NN, FCNYM-UNLP SN) and Nothrotheriops shastensis (New Mexico,

USA, 1 specimen, YPM VP 013198). Hair and diet samples for five modern xenarthran species

(Bradypus variegatus [n=2, three-toed sloth], Choloepus hoffmanii [n= 5, two-toed sloth],

Cyclopes didactylus [n=2, silky anteater], Tamandua tetradactyla [n=5, lesser anteater], and

Myrmecophaga tridactyla [n=2, giant anteater]), kept under controlled feeding conditions, were sampled from the Huachipa Zoo (Lima, Peru) in 2018 and 2019 (Tables 3.1-3.6). Fossil and modern specimens were analyzed for bulk d15N, as well as for d15N amino acid compound specific isotope analysis (AACSIA) at the Department of Geology and Geophysics, University of

Hawaii at Manoa (Figs. 3.1-3.8; Tables 3.1-3.10). To evaluate the homogeneity or heterogeneity of diets over the time represented by a hair, hair strands for all species (modern and fossils) were sectioned in 0.5 or 1 cm sections and analyzed for bulk d15N prior to AACSIA (Fig. 3.2). If not significant differences were found among the different hair sections, entire hair strands then were homogenized and analyzed for AACSIA d15N; otherwise, sections with significantly different bulk d15N were analyzed separately for AACSIA d15N. Diet samples for the modern species were also sampled for bulk and AACSIA d15N. Dietary items for the species with mixed diets were isolated and analyzed for bulk d15N. d15N AACSIA were conducted on the homogenized mixed diets prior to the isotopic analyses, in the proportion of each individual dietary item given to the individual animals at the zoo.

Among the modern species, only the three-toed sloth Bradypus is an obligate herbivore with a monospecific diet (exclusively feeding on young sprouts of Ficus elastica, the rubber plant). The biology of Cyclopes didactylus (silky or pygmy anteater) is not well known in the wild (it is the least studied of xenarthrans), but it is described as a strict insectivore (Best &

Harada 1985). In the zoo, the analyzed specimens are fed a 1:1 formula of two animal protein mixtures (milk replacers, Table 3.5). The other three xenarthran species, the two-toed sloth

Choloepus (an herbivore in the wild) and the anteaters Myrmecophaga and Tamandua

(insectivores in the wild) in the zoo are fed a combination of 8 and 9 items respectively, including items of animal origin. Details for all of the diets are described in Tables 3.2-3.6.

3.2 Amino acid d15N from hair keratin and diets

Hair strands of ten modern specimens, four dietary samples, and four fossil hair samples were analyzed via d15N AACSIA. Thirteen amino acids were consistently measured in all samples: four “source” amino acids (phenylalanine [Phe], glycine [Gly], lysine [Lys], and serine

[Ser]), eight “trophic” amino acids (glutamic acid [Glx, actually combined glutamic acid+glutamine], proline [Pro], alanine [Ala], aspartic acid [Asx], valine [Val], leucine [Leu], and isoleucine [Iso]), and the “metabolic” amino acid threonine (Thr). Nomenclature for

“source”, “trophic”, and “metabolic” amino acids follows (15,16). In a few cases, d15N values for the source amino acids tyrosine (Tyr) and methionine (Met) also were recovered. The metabolic

AA Thr is the most depleted in 15N in all the animal samples, so when describing the lower end of the d15N range of AA values, Thr and the next AA with lowest d15N are included. The analysis of both consumer and diet samples permit calculation of trophic discrimination in d15N from

15 dietary protein to hair keratin for each AA. This is referred to as “big delta” (D Nconsumer-diet).

3.3 Nitrogen isotope analyses of amino acids

Amino acids are chemically isolated and purified from sample material prior to derivatization and esterification in preparation for GC-IRMS analysis. Samples are then analyzed using either a Thermo Scientific Delta V Plus or a Thermo Scientific MAT 253 mass spectrometer interfaced with a Thermo Finnigan Trace GC gas chromatograph via a Thermo

Finnigan GC-C III.

Amino acids measured using this technique include alanine (Ala), glycine (Gly), isoleucine (Iso), leucine (Leu), lysine (Lys), methionine (Met), phenylalanine (Phe), proline

(Pro), serine (Ser), threonine (Thr), tyrosine (Tyr), and valine (Val). Additionally, the terminal amide groups in glutamine (Gln) and aspartamine (Asn) are cleaved during the chemical isolation of amino acids, the result being the conversion of these amino acids to glutamic acid

(Glu) and aspartic acid (Asp), respectively. Thus, the isotope value of a combined Glu + Gln is measured (termed Glx), and the isotope value of a combined Asn + Asp is measured (termed

Asx). Internal reference compounds, L-2-Aminoadipic acid (AAA) and L-(+)-Norleucine (Nor) of known nitrogen isotopic composition, are co-injected with samples and used to normalize the measured isotope values of unknown amino acids and determine accuracy and precision.

Between triplicate runs of each sample, a suite of these 14 amino acids of known isotopic composition are analyzed; AAA and Nor are also co-injected with these amino acids. For isotopic correction of unknown amino acids when co-injected standards are unusable due to co- elution effects, a linear correction is derived from the amino acid suites run immediately before and after the triplicate sample analysis and applied to measured isotope ratios. All amino acids are analyzed in triplicate and isotopic values are reported in δ-notation relative to atmospheric

N2.

3.4 Trophic position estimation and testing

The predictive power of the widely applied equation of (12) to calculate the trophic level of the modern species with known diets first was evaluated. The equation for any studied tissue sampled is described as:

�� = !"#$ %!"#$ %& + 1 (Eq. 1) '()

15 15 15 Where b is the difference between the d N of Glx (d NGlx) and Phe (d NPhe) in the

15 15 ecosystem’s primary producer, and TDF is the offset in d NGlx relative to d NPhe with increasing trophic level, i.e., TDF= (Glxconsumer - Glxdiet) – (Pheconsumer – Phediet). The b and TDF values used were -8.4‰ and 7.6‰ respectively, as described previously (28,121). The trophic level estimated via this equation was then assessed for reliability relative to the known diets/trophic level of the animals sampled.

RESULTS

3.5 Bulk d15N of hair keratin and diets of modern species

Bradypus variegatus (three-toed sloth). Hair strands for the two Bradypus specimens were sectioned in 4 or 5 segments of 0.5 cm each (Table 3.1). The median d15N of Bradypus’ hair samples is 10.9‰ ± 0.3‰, ranging from 10.5‰ to 11.4‰ across the segments and individual hair samples (Fig. 1, Table 1). Differences in bulk d15N for Bradypus’ hair segments within and among specimens are significant (p-value= 0.02 and 0.05 respectively), but within instrument error (0.4-0.6‰). The median d15N of the diet for Bradypus (sampled weekly for 10 weeks in 2016, and once in 2018 and 2019) was 7.1‰ ± 0.7‰ (Table 3.2).

Table 3.1: Bulk d15N of the hair samples for the two specimens of Bradypus variegatus sampled in 2018 and 2019. Zoo ID refers to the identification (chip) number of the specimen at the zoo.

Bulk d15N (‰) Bradypus variegatus Hair Zoo ID: _372836 Hair Zoo ID: _372840 segments 2018 2019 segments 2018 2019 0 – 0.5 cm 11.2 10.8 0 – 0.5 cm 11.1 11 0.5 – 1 cm 10.9 10.7 0.5 – 1 cm 11.1 10.7 1 – 1.5 cm 11.2 10.5 1 – 1.5 cm 11.2 10.7 1.5 – 2 cm 11.4 10.5 1.5 – end 10.8 10.6 2 - end 11.3 NA Median 11.2 10.6 Median 11.1 10.7

Table 3.2: Bulk d15N of the Bradypus’s diet samples (Ficus elastica, rubber plant), sampled in 2016, 2018, and 2019 (ID numbers ending in -16, 2018, or 2019, respectively).

Bradypus’ diet Sample ID d15N (‰) Ficus elastica BRA-DIET2019 8.5 BRA-DIET2018 7.0 BRA-05-01-16 6.1 BRA-06-01-16 7.2 BRA-07-01-16 6.0 BRA-08-01-16 6.8 BRA-09-01-16 7.4 BRA-10-01-16 6.7 BRA-11-01-16 6.9 CHO-04-01-16H 7.0 BRA-05-01-16 7.7 BRA-05-01-16 7.7 Median 7.1 ± 0.7

Choloepus hoffmanii (two-toed sloth). The median d15N of Choloepus hair samples is

7.2‰ ± 1.1‰, ranging from 6.7‰ to 8‰ across the segments and individual hair samples (Fig.

3.1, Table 3.3). The median d15N of the homogenized diet of Choloepus (collected weekly for 7 weeks in 2016, and once in 2019) was 2.5‰ (± 0.2‰), ranging from 2‰ to 2.8‰ (Table 3.4).

Table 3.3: Bulk d15N of the hair samples for the different specimens of Choloepus hoffmanii sampled in 2018 (Cho-01-19 to Cho-05-19) and 2019 (Cho-08-19 to Cho-11-19). Zoo ID refers to the identification (chip) number of the specimen at the zoo.

Bulk d15N (‰) Zoo ID Sample ID Choloepus hoffmanii 2018 2019 373213 (Candy) Cho-01-19, Cho-11-19 8.0 7.6 373191 (Beny) Cho-02-19, Cho-10-19 7.7 7.6 373486 (Sandra) Cho-03-19, Cho-09-19 6.9 6.7 372841 (Speedy) Cho-05-19, Cho-08-19 7.2 7.1 373220 (Lia) Cho-04-19 7.6 - Median 7.2 ± 1.1

Table 3.4: Bulk d15N of Choloepus’ homogenized diet samples (ID numbers ending in - 16 were sampled in 2016 and Di-2019 in that year).

Choloepus’ diet Sample ID d15N (‰) Sweet potato, carrot, broccoli, Cho-Di-2019 2.4 spinach, boiled quinoa, Cho-05-16 2.0 sprouts of rubber plant, Cho-06-16 2.5 Purina® Dog Chow, Cho-07-16 2.4 Aminomix® (protein Cho-08-16 2.8 supplement) Cho-09-16 2.4 Cho-10-16 2.6 Cho-11-16 2.6 Median 2.5 ± 0.2

Cyclopes didactylus (pygmy anteater). Three Cyclopes specimens were sampled in 2019.

The median d15N of Cyclopes’ hair is 8.7‰ ± 0.2‰ (Fig. 3.1, Table 3.5). Cyclopes diet at the zoo is a two-component animal protein milk replacer mixture (wheyTM protein + Esbilac® powder [dietary supplement]). The average d15N for the diet of this species was 5.3‰ ± 0.1‰

(Table 3.5).

Table 3.5: Bulk d15N of hair and diet samples of Cyclopes didactylus.

Hair keratin Diet samples Sample ID d15N (‰) Cyclopes’ diet Sample ID d15N (‰) Cy-01-19 (Paulina) 8.3 Esbilac® powder, whey Cy-Fre-Di19 5.3 Cy-02-19 (Freddy) 8.7 protein (Gold standard 100% Cy-Pa-Di19 TM Cy-03-19 8.7 whey French vanilla flavor) 5.2 Median 8.7 ± 0.2

Tamandua tetradactyla (lesser anteater) and Myrmecophaga tridactyla (giant anteater).

Four Tamandua and two Myrmecophaga individuals were sampled in 2019. The median d15N of

Tamandua’s hair is 5.9‰ ± 0.2‰ and that of Myrmecophaga is 7‰ ± 0.2‰ (Fig. 3.1, Table

3.6). Tamandua and Myrmecophaga’s diet (both are fed the same mixed diet at the zoo) showed a d15N of 5.3‰ (Table 3.6).

Table 3.6: Bulk d15N of the hair and diet samples of Myrmecophaga tridactyla and Tamandua tetradactyla (both anteater species have the same diet).

Hair keratin Diet d15N (‰) Taxon Sample ID d15N (‰) Purina® Cat Chow, soy milk, 5.3 Myrmecophaga Myr-01-19 6.8 boiled egg, carrot, horse meat, Myrmecophaga Myr-02-19 7.1 banana, papaya, Aminomix® Tamandua Ta-01-19 5.8 (protein supplement), Pecutrin® Tamandua Ta-03-19 5.9 (vitamin and mineral supplement) Tamandua Ta-04-19 5.8 Tamandua Ta-05-19 6.1

Figure 3.1: Bulk d15N of the hair samples for the five modern xenarthran species sampled (Bradypus variegatus, Choloepus hoffmanii, Cyclopes didactylus, Tamandua tetradactyla, Myrmecophaga tridactyla), and their diets, as well as the fossil hair samples from Mylodon darwinii and Nothrotheriops shastensis.

3.6 Bulk d15N of hair keratin for the fossil species

Hair samples for two specimens of Mylodon darwinii and one specimen of

Nothrotheriops shastensis were sampled. Hair strands of one individual per species were sectioned in 1 cm increments and analyzed for d15N and molar C:N ratio (Fig. 3.2, Table 3.7).

The C:N molar ratio for Mylodon is 3.6 and that for Nothrotheriops is 3.1 (Table 3.7). These C:N molar ratios indicate that the samples are all well preserved, because they had values within the range expected for extant mammal hair samples (i.e., <4). The d15N value of the proximal 1 cm was higher than the remaining length of the hair for both species and excluded for AACSIA.

Significant differences (p-value< 0.01) were found between Mylodon darwinii’s hair segments 3-

4 plus 7-10 (mean= 2‰ ±0.1), and segments 2 plus 5-6 (mean= 3.5‰ ±0.6). Therefore, I decided to homogenize segments 3-4 plus 7-10 and segments 2 plus 5-6 and to analyze them separately for AACSIA d15N. In contrast, when excluding the first centimeter (d15N=17.5‰), the d15N values of the remaining hair segments of Nothrotheriops are extremely consistent and not significantly different (mean= 16‰ ±0.1). Due to these results, the first segment was excluded for Nothrotheriops and the remaining hair segments homogenized and analyzed for AACSIA d15N.

Figure 3.2: Serial sampling of the hair for the fossil ground sloths Mylodon darwinii (left) and Nothrotheriops shastensis (right). The x-axis represents the sections (cm) of the hair analyzed, 0 on the axis represents the proximal section.

Table 3.7: Bulk d15N (‰ vs. AIR) and C:N ratio (mol/mol) for hair segments of Mylodon darwinii and Nothrotheriops shastensis.

Mylodon darwinii Nothrotheriops shastensis Segments (cm) δ15N C:N Segments (cm) δ15N C:N 0-1 4.3 3.5 0-1 17.5 3.1 1-2 2.9 3.6 1-2 16.1 3.1 2-3 2.1 3.5 2-3 15.9 3.1 3-4 1.9 3.6 3-4 16.0 3.1 4-5 3.4 3.6 5-6 3.4 3.6 6-7 2.1 3.6 7-8 1.9 3.5 8-9 1.9 3.5 9-10 1.8 3.5

3.7 AACSIA d15N of hair keratin and diet samples

Among source AAs in hair keratin, phenylalanine and serine have the highest d15N values for all species, whereas the source AAs with the highest d15N values in dietary components were phenylalanine and glycine (Table 8-9). Among trophic AAs, those with highest d15N values in the hair keratin were proline, isoleucine, and glutamic acid, whereas there was no consistency in the d15N values of trophic dietary AAs for the secondary consumers. The metabolic AA threonine was the most depleted in 15N for all samples and species.

Bradypus variegatus. Hair samples of three Bradypus specimens were analyzed for d15N

AACSIA (Fig. 3, Table 8). Median values of d15N AAs of hair keratin are bracketed by the Thr and Lys values (0.3‰ and 9.7‰ respectively) at the lower extreme and Pro (15.2‰) at the upper end. Dietary AA d15N values range from Leu (6‰) to Phe (18.6‰; Table 3.9). There are no significant differences between the mean d15N of source (12‰) and trophic (13‰) AAs (p- value=0.3) in hair keratin of Bradypus, nor between source (11.3‰) and trophic (9.2‰) AAs (p-

15 15 value>0.1) in the diet of Bradypus. Offsets in d N from diet to hair keratin (D Nconsumer-diet) vary among AAs, ranging from Thr and Phe (-8.5‰ and -4.5‰ respectively) to Pro (6‰).

15 Differences in the D Nconsumer-diet between source (0.6‰) and trophic (4.0‰) AAs are not

15 significant (p-value= 0.2). Of the four source AAs evaluated, only Lys and Gly have D Nconsumer- diet within one standard deviation of ~0‰. In contrast, Phe (-4.5‰ ±1.28) and Ser (5.7‰ ±0.95)

15 have D Nconsumer-diet values even larger than many trophic AAs, including Glx.

Figure 3.3: Median d15N values for individual AA of Bradypus’ hair keratin (red) and its monospecific diet Ficus elastica (teal). Differences between Bradypus’ hair keratin and diet d15N values (D15Nconsumer-diet) for each AA (black) represent d15Nconsumer - d15Ndiet. AAs separated into source (Phe, Gly, Lys, Ser), trophic (Glx, Pro, Ala, Asx, Val, Leu, Iso), and metabolic (Thr). Bulk d15N included for reference. Error bars represent ± 1 standard deviation of multiple samples.) of the hair analyzed, 0 on the axis represents the proximal section.

Choloepus hoffmanii. Hair samples of three individuals of Choloepus were analyzed for d15N AACSIA (Table 3.8). Median values of d15N AAs of Choloepus hair keratin are bracketed by Thr and Tyr (-5.5‰ and -3.8‰ respectively) at the lower end and Pro (11.5‰) at the upper

70 end (Fig. 3.4). Dietary AA d15N ranges from Thr and Ser (0.3‰ and 1.2‰ respectively) to Ala

(8.6‰; Table 9). There are no significant differences between the mean d15N of source (6.3‰) and trophic (9.8‰) AAs (p-value= 0.1) of Choloepus hair keratin, nor between source (3.7‰) and trophic (5.9‰) AAs (p-value> 0.1) of Choloepus’ diet. Offset in d15N from diet to hair

15 keratin (D Nconsumer-diet) range from -5.9‰ (Thr) and -1.7‰ (Gly) to 8.1‰ (Ser) in Choloepus.

15 Differences in mean D Nconsumer-diet between source (2.6‰) and trophic (3.9‰) AAs are not

15 significant (p-value=0.5). Lys is the only source AA with a D Nconsumer-diet value close to 0‰.

Figure 3.4: Median d15N values for individual AA of Choloepus’ hair keratin (red) and its diet (teal). Differences between Choloepus’ hair keratin and diet d15N values (D15Nconsumer- diet) for each AA (black) represent d15Nconsumer - d15Ndiet. AAs separated into source (Phe, Gly, Lys, Ser), trophic (Glx, Pro, Ala, Asx, Val, Leu, Iso), and metabolic (Thr). Bulk d15N included for reference. Error bars represent ± 1 standard deviation of multiple samples.

Cyclopes didactylus. Hair samples of one Cyclopes specimen was analyzed for d15N

AACSIA (Fig. 5). Values of d15N AAs of hair keratin range from Thr and Gly (-9‰ and 3.2‰ respectively) to Iso (15.3‰), while those of its diet range from Tyr and Thr (-1.2‰ and -0.3 respectively) to Phe (8.6‰; Table 3.8-3.9). There are no significant differences between the mean d15N of source (8.2‰) and trophic (12.8‰) AAs (p-value=0.08) in Cyclopes hair keratin, nor between source (4‰) and trophic (6.7‰) AAs (p-value=0.2) of Cyclopes’ diet. Offset in

15 15 d N from diet to hair keratin (D Nconsumer-diet) range from Thr and Gly (-8.7‰ and -3‰

15 respectively) to Pro (9‰) in Cyclopes. Differences in the mean D Nconsumer-diet between source

(4.2‰) and trophic (6.1‰) AAs are not significant (p-value=0.4). All source AAs have

15 D Nconsumer-diet values significantly different from 0‰.

Figure 3.5: Mean d15N values for each AA for Cyclopes’ hair keratin and its mixed diet. Legend as in previous figures. Error bars represent ± 1 standard deviation of multiple samples.

Myrmecophaga tridactyla and Tamandua tetradactyla. Hair samples of Myrmecophaga and Tamandua specimens were analyzed for d15N AACSIA (Fig. 3.6). Values of d15N AAs of hair keratin range from Thr and Tyr from both Myrmecophaga (-13‰ and -2.3‰ respectively) and Tamandua (-12‰ and 1.8‰ respectively) to Pro (15.3‰ in Myrmecophaga and 14.4‰ in

Tamandua). The d15N of dietary AAs range from Thr and Lys (-1.1‰ and 2.8‰) to Ala (11.6‰,

Table 9) in these two species. Differences between the mean d15N of source (4.5‰) and trophic

(10.1‰) AAs of Tamandua and Myrmecophaga hair keratin are significant (p-value<0.01).

However, the d15N of dietary source (7.2‰) and trophic (8.6‰) AAs do not differ significantly

15 (p-value=0.5). Differences in the mean D Nconsumer-diet between source (-1.4‰) and trophic

(1.5‰) AAs also are not significant (p-value=0.06). Among source AAs, Ser and Lys have

15 D Nconsumer-diet values within one standard deviation of ~0‰.

Figure 3.6: Mean d15N values for each AA for Myrmecophaga and Tamanadua’s hair keratin and diet (symbols linked with lines; both species are fed the same diet). Symbols without lines represent the D15Nconsumer-diet for each species. Error bars represent ± 1 standard deviation of multiple samples.

Mylodon darwinii and Nothrotheriops shastensis. The differences in the AA d15N between these two species of fossil sloths are markedly dissimilar and significant (p-value<

0.001). The three samples of Mylodon show consistent patterns of d15N AACSIA (Fig. 3.7), with median values ranging from Thr and Tyr (-7.2‰ and -3.4‰ respectively) to Iso and Pro (5.9‰).

The largest d15N variation is observed in Ala, with almost 5‰ variation between the two individuals of Mylodon sampled. There are significant differences (p-value<0.001) among source

(mean d15N= 0.78‰) and trophic AAs (mean d15N= 4.92‰) of hair keratin in Mylodon. 74

Nothrotheriops AA d15N values vary widely (15‰), ranging from Thr and Lys (4.5‰ and 9.8‰ respectively) to Glx (19.6‰). The AA d15N values of Nothrotheriops are the most enriched in

15N of all the species analyzed. Differences between the mean of source (14.1‰) and trophic

(17.13‰) AAs of Nothrotheriops hair keratin are not significant (p-value= 0.1).

Figure 3.7: Mean AA d15N values for hair keratin of Nothrotheriops shastensis and three samples (2 individuals) of Mylodon darwinii.

3.8 d15N of glutamic acid and phenylalanine

The seven species of modern and fossil xenarthrans in this study show substantial intra-

15 and inter-specific variation in their d NPhe values (>15‰ across individuals), with the spectrum

15 of d NPhe variation being bracketed by the two fossil sloth species, Mylodon (median

15 15 15 d NPhe=2.5‰) and Nothrotheriops (d NPhe=17.7‰). The greatest intraspecific d NPhe variation

15 is observed in the omnivorous Choloepus (8.4‰ span), with hair keratin d NPhe values ranging

15 15 from 3.2‰ to 11.6‰. Intraspecific variation in d NGlx values is smaller than that for d NPhe, with Choloepus and Mylodon showing the greatest variation among individuals (2.5‰).

15 Interspecific variation in d NGlx spans 14‰, with the extreme values also bracketed by the fossil

15 15 sloth species Mylodon (median d NGlx= 5.6‰) and Nothrotheriops (d NGlx= 19.6‰).

15 15 Variation for both d NPhe and d NGlx values for the four different diet samples (i.e., diets for Choloepus, Bradypus, Cyclopes, and Tamandua/Myrmecophaga) are bracketed by the diets

15 15 of the two modern sloth species, Choloepus (d NPhe= 4.5‰, d NGlx= 5.3‰) and Bradypus

15 15 15 (d NPhe= 18.6‰, d NGlx= 11‰; Table 3.9). The largest offset between dietary d NGlx and

15 d NPhe (i.e., the b value) is observed in Bradypus (-7.6‰), a species with a monospecific herbivorous diet (the rubber plant Ficus elastica). The species with mixed omnivorous diets show small variations in their b values (<2‰). The b value in the pygmy anteater Cyclopes is -

1‰, 0.75‰ in Choloepus, and -0.37‰ in Myrmecophaga and Tamandua (both species fed the same diet at the zoo).

Table 3.8: AA and bulk tissue δ15N values for modern and fossil (†) xenarthran species. TL is the trophic level calculated by Eq. 1, using the previously published values of β = -8.4‰ and TEF = 7.6‰.

d15N (‰) Sample TL Ala Gly Thr Ser Val Leu Iso Pro Asx Glx Phe Tyr Lys Bulk 12 11.9 0.28 14 14.7 11.5 14.8 15.2 12 14.6 15.3 11.9 9.74 10.5 2.0 Bradypus 12.3 10.1 -1 13 13.3 11.9 14.2 15.8 11.6 14.5 14.1 11.6 11.1 10.6 2.2 variegatus 9.69 11.3 0.74 13 12.8 11.7 10.4 14.9 11.3 14.1 12.9 7.81 6.57 10.8 2.3 Median 12 11.3 0.28 13 13.3 11.7 14.2 15.2 11.6 14.5 14.1 11.6 9.74 Bradypus 7.79 5.14 -7.6 9.3 8.97 8.03 9.72 11.1 8.3 11.2 3.24 3.05 2.41 7.1 3.2 Choloepus 9.24 4.26 -5.5 8.9 9.33 8.11 10.8 11.5 8.59 11.5 11.6 3.76 2.28 7.6 2.1 hoffmanii 11.5 6.66 -5.3 10 8.48 10.6 12.4 13.5 12.1 13.7 8.17 6.11 5.71 7.6 2.8 Median 9.24 5.14 -5.5 9.3 8.97 8.11 10.8 11.5 8.59 11.5 8.17 3.76 2.41 Choloepus Cyclopes didactylus 10.8 3.2 -9 9.3 11.3 11.8 15.3 14.7 10.9 14.5 14.8 7.72 5.78 8.3 2.1

Tamandua 10.1 1.54 -12 6 8.06 6.71 12.1 14.9 7.58 10.2 5.3 2.35 2.74 5.8 2.7 tetradactyla 8.94 3.14 -12 7.8 8.23 7.88 11.1 13.9 7.3 9.74 10.4 1.17 2.1 5.8 2.0 Average 9.52 2.34 -12 6.9 8.14 7.29 11.6 14.4 7.44 9.95 7.86 1.76 2.42 Tamandua Myrmecophaga 10.5 3.71 -13 8.7 8.25 8.81 10.6 15.3 9.1 11 12.1 -2.3 1.96 7.1 2.0 tridactyla 1.7 -0.4 -7.2 4.9 4.2 1.1 3.3 3.8 3.1 4.6 2 -4 -2.5 2 2.4 †Mylodon 2.9 1.5 -5.7 5.5 7.6 3.3 5.9 5.9 4.5 5.6 2.5 -3.4 -3.6 3.4 2.5 darwinii 6.6 2.2 -7.8 5.0 8.1 4.5 6.7 6.9 5.9 7.1 4.4 -1.3 -1.1 3.7 2.5 Median 2.9 1.5 -7.2 5.0 7.6 3.3 5.9 5.9 4.5 5.6 2.5 -3.4 -2.5 Mylodon †Nothrotheriops shastensis 15.6 13.3 4.5 17 16.7 15.9 17.9 18.4 15.8 19.6 17.7 12.8 9.8 16.4 2.4

Table 3.9: AA and bulk tissue δ15N values for dietary samples of the modern xenarthran species included in this study. TL as above. Δ15NConsumer–Diet= δ15NConsumer – δ15NDiet. δ15NConsumer from previous Table.

d15N (‰) Sample TL Ala Gly Thr Ser Val Leu Iso Pro Asx Glx Phe Tyr Lys Bulk Bradypus’ diet 10.9 11.9 8.8 7 9.27 5.98 8.37 9.2 9.44 11 18.6 7.8 8.5 1.1 (Ficus elastica) ∆consumer-diet 1.11 -0.6 -8.5 5.7 4 5.72 5.78 6.01 2.14 3.54 -4.5 1.9

Choloepus’ diet 8.55 6.82 0.3 1.2 4.37 2.39 5.08 8.58 7.08 5.28 4.53 2.2 2.4 2.2 (mixed) ∆consumer-diet 0.69 -1.7 -5.9 8.1 4.61 5.71 5.71 2.96 1.5 6.21 3.64 0.2

Cyclopes’ diet (mixed of 2 7.03 6.23 -0.3 5.3 8.09 3.98 7.26 5.7 6.97 7.57 8.57 -1.2 1.1 5.2 2 items) ∆consumer-diet 3.76 -3 -8.7 4 3.19 7.83 8.05 8.98 3.9 6.9 6.24 8.94 4.7

Tamandua Myrmecophaga’s 11.6 9.69 -1.1 6.2 6.61 5.68 8.63 9 9.13 9.49 9.86 2.8 5.3 2.1 diet (mixed) ∆consumer-diet -1.1 -6 -12 2.5 1.64 3.13 1.96 6.31 -0 1.5 2.27 -0.9 Myrmecophaga ∆consumer-diet -2.1 -7.3 -11 0.7 1.53 1.62 2.97 5.42 -1.7 0.47 -2 -0.4 Tamandua

3.9 Trophic position estimations for modern and fossil xenarthrans using constant b and TDF values

Trophic positions for all samples were obtained using the trophic position (TP) equation described above (Eq. 1) with the constant b (-8.4‰) and TDF (7.6‰) values proposed by

Chikaraishi et al. (2011). In this equation, a TP=1 is expected for a primary producer, TP= 2 for a primary consumer, TP=3 for a secondary consumer, and TP= 4 would represent a hypercarnivore. This equation predicted the trophic position of Bradypus as a herbivore (average

TP= 2.1, Table 8). The TPs obtained for the three specimens of Choloepus varied from 2.1 to 3.2

(Table 8), i.e., interpretable as either a pure herbivore or a secondary consumer. Cyclopes, a species with a mixed diet that includes animal whey protein (protein isolates from cow’s milk) showed a TP of 2.1, which would correspond to that expected for a primary consumer. The trophic positions obtained for the anteaters Tamandua and Myrmecophaga (whose diets at the zoo include horse meat and other elements of animal origin) varied from 2.0 for Myrmecophaga and one Tamandua specimen, to 2.7 for the second individual of Tamandua, reflecting a trophic level interpretation as a primary consumer to intermediate between primary and secondary consumer (Table 8). The TP obtained for both fossil sloth species, Mylodon and Nothrotheriops, was 2.5 and this value was consistent among individuals (Table 3.8). Eq. 1 also was applied to the dietary samples (Table 3.9). The TP obtained for Ficus elastica (the exclusive dietary source for Bradypus) was 1.1, consistent with what would be expected for a primary producer (Table

3.9). The TPs obtained for the three mixed diets range from 2 to 2.2 (Table 3.9).

Table 3.10: Summary of the b (d15NGlx-d15NPhe in the diets), TDF, and ∆d15NGlx-d15NPhe values for all species in this study.

Bradypus Choloepus Cyclopes Myrmecophaga Tamandua Species variegatus hoffmanii didactylus tridactyla tetradactyla Diet Ficus elastica mixed diet mixed diet mixed diet mixed diet b value -7.61 0.75 -1.0 -0.37 -0.37 Consumer 15 15 0.44 3.32 -0.34 -1.14 2.09 ∆d NGlx-d NPhe TDF 8.05 2.57 0.66 -0.77 2.46

DISCUSSION

3.10 Assumptions behind the equation to calculate trophic positions using AA d15N values

The widely used equation to calculate trophic positions using AA d15N (Eq. 1) relies on

15 15 the following three premises. First that the offset between the d N of glutamic acid (d NGlx) and

15 phenylalanine (d NPhe) in the organism under study will increase with increasing trophic level.

15 15 Second, that the b value (the offset between d NGlx and d NPhe at the base of the food chain, i.e., the primary producers) is relatively constant and has a value of -8.4‰ for C3 ecosystems, 0.4‰ for C4 ecosystems, and 3.4‰ for marine food webs. Third, that the trophic discrimination factor

15 15 in all ecosystems (TDF, the difference in the d NGlx relative to d NPhe with increasing trophic level) is 7.6‰.

15 Concerning the first assumption, empirical data show that d NPhe values in a consumer

15 15 remain relatively unchanged in relation to its ingested dietary d NPhe, whereas the d NGlx in a

15 consumer experiences substantial fractionation relative to the dietary d NGlx (28,120,121,124).

Explanations behind these observations are twofold. First, during metabolic processes, phenylalanine (as a “source” AA) seldom forms nor breaks carbon-nitrogen bonds (124). Indeed, the main metabolic route of phenylalanine (leading to the formation of tyrosine) does not involve

15 a nitrogen-associated reaction, and therefore, it does not lead to d NPhe discrimination (11). In

79 contrast, trophic AAs are involved in transamination (transfer of amino groups to form a keto acid and another AA) and deamination (removal of an amino group from an AA) reactions, which would inevitably result in isotopic discrimination. It follows that the higher the position of

15 15 an organism in the trophic level the greater the expected offset between d NGlx and d NPhe. AAs with readily interchangeable a-nitrogens (i.e, trophic AAs) are linked via glutamic acid, which is crucial in nitrogen metabolism (15). Indeed, glutamic acid is involved in the transamination of many AAs, and is how the amino acid nitrogen enters the urea cycle (15,125).

Although averaging the d15N of source and trophic AAs does not result in significant differences in most taxa (with the exception of Tamandua and Mylodon, Table 11), results here indicate that glutamic acid (together with proline and isoleucine) indeed shows higher d15N values than source and other trophic AAs, as expected (Fig. 3.3-3.7, Table 3.8). Furthermore, the

15 15 results herein document that the ∆d NGlx-d NPhe, which is the first premise in which Eq. 1 is

15 15 based, also are as expected overall. Indeed, ∆d NGlx-d NPhe values are lower in the primary consumers Bradypus (0.4‰) and higher in the secondary consumers Tamandua and Choloepus

(2.1‰ and 3.3‰ respectively, Table 3.10). Results for the obligate folivore Bradypus and the mixed feeders Tamandua and Choloepus are consistent with data from other mammalian species.

Indeed, when these results are viewed within a broader mammalian context (data exists for other

15 15 24 other mammalian species), herbivores show significantly lower ∆d NGlx-d NPhe values

15 15 15 15 (n=11, mean ∆d NGlx-d NPhe= -0.6‰ ± 0.6) than mixed feeders (n=5, mean ∆d NGlx-d NPhe=

15 15 3.2‰ ± 1.6; p-value= 0.02; Figure 3.8). However, the ∆d NGlx-d NPhe values obtained for

Myrmecophaga (-1.1‰) and Cyclopes (-0.3‰) are lower than expected, with values corresponding to that of herbivorous species, rather than to mixed feeders. For Cyclopes, this can be attributed to its artificial diet at the zoo. Indeed, although one of the two protein mixtures of

Cyclopes’ diet includes whey protein (which is generally derived from cow’s milk, an animal protein that should result in isotopic values indicating a secondary consumer), it is possible that this animal protein element represents only a small proportion relative to the other plant-derived ingredients in this protein mixture. A potential explanation for low values in Myrmecophaga is more difficult to find, because it is fed the same diet as Tamandua. Assuming that these isotopic data are representative for the species (results for both taxa are based on one specimen, although that also is the case for most mammal species in the literature), it shows either unrecognized uncertainties inherent in the variables of the base equation or peculiarities behind AA synthesis in these anteaters (including those that might be associated with a mixed source diet differing

15 substantially from their exclusively insect-based diet in the wild). That the d NPhe of the

15 anteaters Myrmecophaga and Cyclopes is higher than d NGlx is contrary to expectations (Figures

3.5-3.6), but more data are necessary before attempting to find broader mechanistic explanations

15 15 behind these results. The ∆d NGlx-d NPhe values obtained for secondary consumers in marine

15 15 ecosystems are highly consistent (n=4, mean ∆d NGlx-d NPhe= 13.1‰, p-value< 0.001), and significantly higher than for any terrestrial feeding guild (Figure 3.8). Data for terrestrial

15 15 carnivores is limited to two species of top predators with very different ∆d NGlx-d NPhe values:

15 15 15 15 Crocuta crocuta (∆d NGlx-d NPhe= 11.2‰) and Canis lupus (∆d NGlx-d NPhe= 5.9‰).

15 15 The second premise involves the offset between d NGlx and d NPhe (b value) at the base of the food chain. In terrestrial environments, the b value was originally defined as -8.4‰ for C3 photosynthesis ecosystems (the average for 12 plant species with b values ranging from -5.2‰ to

-12.1‰) and 0.4‰ for C4 ecosystems (the average for 5 plant species with b values ranging from -2.8‰ to 2.9‰; Chikaraishi et al 2010). Although obtained from a single ecosystem (a farm in Japan) these b values are assumed to be representative of C3 and C4 plants globally.

Although other studies obtained similar results, significantly different b values were found in other studies of terrestrial plants ranging from -10.7‰ to -3.3‰ (123,126,127). In fact, both

(127) and (126) analyzed the same species of wheat (Triticus aestivum) but obtained average b values that differ significantly (-4.2‰ vs. -7.55‰). The results of my controlled feeding experiments also show that the b value is not constant, as previously observed in other AACSIA studies, and that a standard -8.4‰ value might not be applicable across taxa. Indeed, the b value of Ficus elastica (rubber plant) is -7.6‰. The b values calculated from the mixed diets in this study (i.e., diets of the two-toed sloth and the three anteater species) are lower, ranging from -

1‰ to 0.75‰. One explanation for the significantly low b values for the mixed diets (despite the fact that most dietary items are of C3 plant origin; Table 3.4-3.6) is the presence of one dietary component of C4 origin. Indeed, the b value for C4 ecosystems has been defined as 0.4‰ (28), which is closer in value to those found in the mixed feeder species included in this study.

Choloepus, Tamandua, and Myrmecophaga have Purina® animal feed in their diets which contain corn and meat (beef and chicken are usually fed corn in the US; Table 3.4-3.5).

However, Purina® represents less than 10% of these animals’ diets, so the low b values in these

15 15 species cannot entirely explain the small offset between the d NGlx and d NPhe values.

Tamandua and Myrmecophaga also have horse meat in their diets, which are usually fed C4 plant species, although that has not been evaluated directly. Similarly, the diet of Cyclopes includes proteins derived from animals (whey isolates derived from cow or goat milk), which are usually fed C4 diets in North America (the U.S. is the source of both products in the diet of

Cyclopes). Because the b values proposed in Eq. 1 only work for well differentiated C3, C4, or marine basal resources, this result highlights an issue that must be considered when applying the

TP equation to study food webs that include a combination of resources.

Finally, the standard TDF value of 7.6‰ (used in Eq. 1) is based only on empirical data from a large number of aquatic organisms (28). A meta-analysis on 359 marine species, however, found a significantly lower TDF value, and suggested that a 7.6‰ value might not be applicable across taxa and ecosystems (122). Values close to the proposed ‘universal’ TDF of

7.6‰ (within 1‰ difference) were also found in insects, fungi, and bacteria (121,128). The results of my controlled feeding study show large variations in TDF across species. The three- toed sloth Bradypus, shows a TDF of 8.1‰, which is within instrumental error of the 7.6‰ value used in the TP equation. The species with mixed diets in this study (the two-toed sloth and the three anteater species), however, show much lower TDF values ranging from -0.8‰ to 2.6‰

(Table 3.10). Another controlled feeding experiment on a mammalian herbivore fed a monospecific diet (the cow Bos taurus) showed a TDF value of 3.9‰ (123), significantly lower than the 7.6‰. Moreover, changes in TDF values might not necessarily reflect increases in the trophic position of an organism, but also might result from variation in the dietary quality, temperature, and nitrogen metabolism (122,129–132).

3.11 Trophic position estimations

The results from applying Eq. 1 to the five species analyzed herein accurately reconstructed the trophic position of Bradypus as an herbivore (TP= 2.1), as well as Choloepus

(TP=2.5) and Tamandua (TP=2.4) as mixed feeders. However, the TPs obtained for the other species are variable and less reliable. The reconstructed trophic position for the mixed-feeders in this study range from that expected for a pure herbivore (i.e., TP= 2) to that expected for a pure carnivore (i.e., TP= 3). The four specimens of fossil sloths consistently yielded TPs of 2.5, i.e., intermediate between a pure herbivore, and a pure carnivore.

Indeed, although it could be argued that a TP=2.5 would be that expected for a species incorporating both plants and meat in its diet (e.g., the value obtained for Choloepus and

Tamandua, known to feed on both in these controlled diet experiments), the interpretation of intermediate values in extinct taxa is not well defined. A TP=2.7 obtained from Homo neandertalensis (Spy Cave, Belgium) was interpreted as this species relying mostly on meat

(~80%) from terrestrial herbivores (Naito et al. 2016), although that interpretation has been contested on the basis of the assumptions behind the constants used (133). The TP values obtained from the three specimens of Mylodon darwinii all are 2.5. Likewise, the TP of the single Nothrotheriops’ specimen is 2.4. These TPs each would be interpreted as representing mixed feeding omnivory, with both plant and meat components in the diet. Eq.1 did not accurately reconstruct the TP of the mixed feeders Myrmecophaga (TP=2) and Cyclopes

(TP=2.1). A possible explanation for these discrepancies relies on the different b values defined for distinct ecosystems. These two species are fed a combination of elements of C3 and C4 origin which might explain the inconsistent results obtained for these omnivore species. It is intriguing, however, that the TPs obtained for Myrmecophaga and Tamandua are so different from each other, given that both species are fed the same diet at the zoo.

15 15 3.12 Correlation between d NGlx and d NPhe

Rather than relying on an equation with constants introducing uncertainties and that are not applicable to organisms feeding on a combination of items of different origin (e.g., C3 + C4

15 15 plants), d NGlx and d NPhe values by themselves can accurately reconstruct the trophic position

15 15 of organisms (Fig. 3.8). When plotted, the results on d NGlx and d NPhe herein obtained for five xenarthran species in controlled feeding experiments, combined with all mammalian data available from the literature, I found strong and significant correlations between these two AAs

15 and with trophic positions. There is a strong significant and positive correlation in the d NGlx

15 2 2 and d NPhe of herbivores (R = 0.98, p-value< 0.001) and marine consumers (R = 0.86, p-value=

0.02). Data for terrestrial carnivores are limited to two analyses, so it is not yet possible to evaluate a correlation with d15N values of these two AAs for this dietary category. However, the

15 15 d NGlx and d NPhe values for these two species are higher than that of herbivores and mixed feeders (Figure 8). Data for mixed-feeders/omnivores are also limited to the four xenarthrans species herein reported plus two Homo species (Homo sapiens and H. neandertalensis). Based

15 15 on these 7 omnivore specimens, no correlation was found in d NGlx and d NPhe for this dietary

15 15 category. Although the d NGlx and d NPhe values of Cyclopes and Myrmecophaga fall within the error associated with the herbivore regression line, more omnivore data are needed to

15 15 evaluate the interspecific variation in d NGlx and d NPhe for this dietary category.

Ordinary least square (OLS) analyses were carried out classifying fossil sloths either as herbivores or as omnivores, to evaluate the best dietary category fit for these species relative to the known correlations in extant taxa in those feeding categories. When the OLS analysis is run

15 15 with fossil sloths classified as herbivores, the correlation between d NGlx and d NPhe is still strong and significant (p-value< 0.001), but the R2 decreases from 0.98 to 0.89. Furthermore, when classified as herbivores, diagnostic analyses (e.g., residual vs leverage) found one of the

Mylodon specimens and Nothrotheriops as outliers, although only Nothrotheriops influenced the

15 15 15 d NGlx and d NPhe regression for herbivores. The exclusion of Nothrotheriops from the d NGlx

15 2 and d NPhe regression further decreased the strength of the correlation to R = 0.86, although the herbivore correlation remained significant (p-value< 0.001). When the fossil sloths were

15 15 classified as omnivores, the correlation between d NGlx and d NPhe for this dietary category transformed from non-existent to strong and significant (R2= 0.86, p-value< 0.001). Results from

85 a robust regression and diagnostic analyses found Myrmecophaga, Homo sapiens with C3 plant diets, and H. neandertalensis as outliers, but none of them were influential for the regression.

The fact that both Mylodon and Nothrotheriops were found as outliers when classified as herbivores, and none of these four specimens were found as outliers when classified as mixed feeders/omnivores, indicates that these fossil taxa best fit within the mixed feeder/omnivore dietary category.

Figure 3.8: Correlations between d15NPhe and d15NGlx for different mammalian species. There is a significant strong positive correlation for mammalian herbivores (green line, R2= 0.98, p-value< 0.001) and mammals feeding on marine resources (blue line, R2= 0.86, p-value= 0.02). No correlation was found for mixed feeders/omnivores; Choloepus and the anteaters Cyclopes, Myrmecophaga, and Tamandua are categorized as mixed-feeders/omnivores based on mixed diet feeding, and known animal and plant source components, in zoo samples analyzed. Fossil sloths excluded from any dietary category, to permit independent assessment of their diets (exclusively herbivorous, as typically presumed, or potentially with secondary consumer components [meat/carnivory]).

3.13 Comparison of d15N AACSIA results to interpretations from other proxies

Numerous studies have shown the strong correlation between dental occlusal surface area

(OSA) in modern herbivores with efficiency and capacity of food processing, digestive physiology, and food quality (134–136). Evaluation of the dental occlusal surface area of xenarthrans relative to modern herbivores showed that six of seven fossil sloth species analyzed had smaller occlusal areas than expected for their body masses (24), indicating that fossil sloths had a lower capacity for food processing than modern herbivores of similar sizes (with the exception of Megatherium americanum which had a higher OSA than even modern herbivores of similar sizes). Of all the fossil sloth species analyzed, mylodontids (and Mylodon darwinii in particular) had the lowest OSA values, reflecting poor oral food processing efficiency. Although metabolisms of fossil sloths are not known, these results were interpreted as likely due to lower energetic requirements of fossil sloths, given the low basal metabolic rates of modern sloths. As such, it was proposed that fossil sloths might have required lower food intake than other mammals of similar sizes, and indirectly argued that they might have had even higher digestive efficiencies than modern ruminants (24).

An alternative explanation for the smaller than predicted OSA of fossil sloths would be a higher food quality ingested. Indeed, if fossil sloths were opportunistic scavengers and incorporated, even if only sporadically, items other than plants in their diets, then the small OSA of fossil sloths would not be unexpected. This interpretation would be in agreement with my results on AACSIA, which clearly indicate that fossil sloths were not obligate herbivores and consumed protein sources of animal origin.

3.14 Conclusions

This study documents the usefulness, power, and potential of AACSIA to predict trophic position of mammals, especially when compared to analyses using only bulk d15N or from indirect inferences based on morphology. Indeed, for the results of the fossil sloths Mylodon and

Nothrotheriops, their bulk d15N values differ by 12‰-14‰, which would represent a difference of approximately four trophic levels if the baseline d15N was the same for both species. The results for the same specimens obtained from AACSIA, however, indicate that these extremely different values were due to differences in the d15N of the baseline rather than different trophic

15 15 levels. Both the TP equation and the regression analyses of d NGlx and d NPhe show that

Mylodon and Nothrotheriops occupy the same trophic position. Importantly, both analyses suggest that the trophic position occupied by these two fossil sloth species were not that of pure herbivores as commonly presumed, but rather that they were both mixed feeders/omnivores, incorporating items of animal origin in their diets. Although the craniodental morphology of sloths precludes them from being active predators, opportunistic scavenging behaviors would be in agreement with the results obtained via AACSIA. Studies on masticatory muscle structure reconstructions, biomechanics, and morphogeometry for both Mylodon and Nothrotheriops indicate that these would have been inefficient at processing foodstuff and in disadvantage (in terms of resource competition) relative to other mammalian herbivores (137,138). Furthermore, morphology-based dietary interpretations of pure herbivory can only be explain with the attribution that fossil sloths had very low metabolic rates (138). This attribution, only supported by the fact that the two living species have low metabolisms, is in disagreement with known ecologies of some extinct sloth species (e.g., that of the semiaquatic sloth Thalassocnus). These results, the first direct evidence of carnivory for ancient sloths diets, should be assessed further

15 15 after obtaining more d NGlx and d NPhe data from extant omnivorous mammal species. Indeed, although there is no single regression that fits the limited current data from extant omnivores, it

15 is certain that there is animal protein source in the diets of all these species and that their d NGlx

15 15 15 and d NPhe regression line falls above that of pure herbivores. More d NGlx and d NPhe data on secondary consumers could allow differentiation of insect- and meat-eating omnivores, as well

15 15 as hypercarnivores as well as teasing apart the interspecific variation in d NGlx and d NPhe relative to particular food sources for mixed feeders.

15 15 More broadly, this study further documents that d NGlx and d NPhe by themselves can be used as predictors for trophic positions. The correlation between these two AA d15N values should be used instead of (or in addition to) relying on an equation that uses fixed values that are unlikely to be universal across distinct ecosystems and that incorporate uncertainties in the trophic position estimations within any particular type of ecosystem. Furthermore, because the b values proposed in Eq. 1 (12,28) for reconstructing trophic position (TP) only work for well differentiated resources (either C3, C4, or marine ecosystems but not a combination of any of these), it precludes the use of this isotopic technique on organisms with diets of different origins.

15 15 Use of d NGlx and d NPhe alone circumvents the limitation of the TP Equation 1 when dealing with food webs that include a combination of resources, like those potentially pertaining to fossil sloths or in early humans, and should be widely applied in future studies of the paleoecology of terrestrial taxa ecosystems.

Epilogue

Stable isotope analyses represent powerful tools to reconstruct ancient ecologies and to reveal ecological and behavioral aspects of modern and extinct organisms that otherwise would likely remain unknown. The techniques, however, are based on assumptions established on specific organisms and ecosystems that might not be transferable across taxa, space and time. By applying stable isotope analyses to a little-studied (but highly diverse and ecologically significant in the fossil record) group of mammals, sloths, and by studying a setting generally neglected in terms of isotopic studies, closed canopy rainforests, my research recognized flaws in some of the principal assumptions behind stable isotope techniques. These results show that the assumptions needed to be investigated and tested in order to confidently interpret stable isotopic signatures of both modern and fossil mammals, and those from different ecosystems. Thus, my research determined that there is not a single diet-bioapatite enrichment (e*diet-bioapatite) factor for all mammalian herbivores, as had been assumed for more than two decades. Based on a previously unexpected but strong and significant correlation with body mass across mammalian herbivores,

I proposed equations that use this parameter to calculate the diet-bioapatite value for the specific herbivorous mammal under study. My results further showed that e*diet-bioapatite is higher in foregut fermenters compared to hindgut fermenters, which is consistent with the significant differences of CH4 production observed in foregut relative to hindgut fermenters. Since CH4 production as the main driver of e*diet-bioapatite differences observed in mammals remains only theoretical, a future direction for this research includes direct experimental measurement of methane across a wide diversity of mammalian herbivores and the evaluation of its potential correlation with e*diet-bioapatite.

Another important assumption, also related to constant factors previously used to make ecological reconstructions, involves the two constants used to predict trophic positions of organisms based on d15N values of individual amino acids, constants that might not be applicable across taxa and ecosystems and whose broader applicability had never been tested empirically before. My results documented that the b value, i.e., the offset between the d15N values of

15 15 glutamic acid (d NGlx) and phenylalanine (d NPhe) in primary producers, proposed as -8.4‰ for

C3 ecosystems, is not universal. Moreover, application of this constant clouds trophic position estimations, because it precludes organisms feeding on a combination of resources of different origin (e.g., C3+C4 plants). Likewise, the trophic discrimination factor or TDF (i.e., the

15 15 difference in d NGlx relative to d NPhe with increasing trophic), assumed to be 7.6‰ for all species and ecosystems, also is not consistent across mammals. Furthermore, the role of this

TDF value deserves further examination, as it also may be influenced by factors other than trophic position, like quality of diet, temperature, and nitrogen metabolism. Moreover, my results show that assuming these constant values is not necessary, and that the correlation between

15 15 d NGlx and d NPhe alone is a reliable predictor of trophic level. There is significant evidence for

15 15 a strong correlation between d NGlx and d NPhe on mammalian herbivores and mammals feeding on marine resources. Although limited, the available and my new data now suggest that that might also be the case for terrestrial secondary consumers. Future directions for this

15 15 research, therefore, involve the evaluation of d NGlx and d NPhe values on secondary consumers

(both hypercarnivores and mixed feeders/omnivores) to be able to determine correlations between these two amino acids within each of these dietary categories, particularly since my results when including mixed feeding extinct sloths suggest that mixed feeders/omnivores do show a significant covariance in the d15N values for these two amino acids.

Finally, by comprehensively analyzing an Amazonian mammalian community, this dissertation also alters current paradigms governing the way closed canopy rainforests are identified using mammalian d13C data. Indeed, the absence in modern Amazonian mammal

13 communities of highly negative d Cdiet values previously proposed as indicative of rainforests

(<-31‰) suggest that the absence of mammals with these values are by no means evidence of absence for closed canopy rainforests. Comparison with analogous data from Africa shows,

13 however, that despite their geographical and taxonomic differences, median d Cdiet values from closed canopy rainforests in Amazonia (-27.4‰) and equatorial Africa (-26.9‰) are not significantly different. Yet, the isotopic range occupied by mammalian herbivores there is narrower than that of equatorial Africa, even though the sampled Amazonian localities span a

13 wider latitudinal range and Amazon closed-canopy rainforest vegetation exhibits a similar d C range to that observed for African plants. I hypothesize that Pleistocene extinctions might have affected niche occupation of modern Amazonian mammals differently from African faunas.

Thus, future directions for this research include the isotopic analyses of pre-Holocene

Amazonian mammals, to test the potential occupation by recently extinct taxa of isotopic niches currently empty in Amazonian mammalian ecosystems, but that are occupied in equatorial

African mammalian communities.

My research program aims to use stable isotope techniques to improve understanding of the evolution of terrestrial ecosystems, biotic interactions, role of extinctions and animal physiology on ecosystem structure, and the ecological and physiological versatility of South

American mammals, including a distinctive group of mammals, sloths. By working in South

American faunas, especially in an understudy setting in terms of paleontological and isotopic studies, such as the Amazon rainforest, I aim to provide a new perspective on the evolution of

South American mammals as a whole, and on the structure of mammalian communities in the tropics. There are, however, limitations in the state of current knowledge that need to be addressed to critically apply the stable isotope techniques to all sorts of taxa and ecosystems.

Thus, the guiding thread of my research includes the testing of foundational cornerstones on which these methods rely, in order to validate the suitability of applying these techniques to different mammalian clades and ecosystems, and outside of South America. Therefore, my research program aims to contribute in the development of the field to more reliably and confidently interpret the isotopic signals preserved in extant and extinct organisms.

References

1. Cerling TE, Hart JA, Hart TB. Stable isotope ecology in the Ituri Forest. Oecologia.

2004;138(1):5–12.

2. DeNiro MJ, Epstein S. Influence of diet on the distribution of carbon isotopes in animals.

Vol. 42, Geochim. Cosmochim. Acta. 1978. p. 495–506.

3. Ambrose SH, Norr L. Experimental evidence for the relationship of the carbon isotope

ratios of whole diet to bone collagen and carbonate. In: Lambert JB, Grupe G, editors.

Prehistoric Human Bone. Springer-Verlag; 1993. p. 1–37.

4. Tieszen LL, Fagre T. Effect of diet quality and composition on the isotopic composition of

respiratory CO2, bone collagen, bioapatite, and soft tissues. In: Lambert JB, Grupe G,

editors. Prehistoric Human Bone. Springer-Verlag; 1993. p. 121–55.

5. Passey BH, Robinson TF, Ayliffe LK, Cerling TE, Sphonheimer M, Dearing MD, et al.

Carbon isotope fractionation between diet, breath CO2, and bioapatite in different

mammals. J Archaeol Sci. 2005;32(10):1459–70.

6. Hautier L, Billet G, Eastwood B, Lane J. Patterns of morphological variation of extant

sloth skulls and their implication for future conservation efforts. Anat Rec.

2014;297(6):979–1008.

7. Vendl C, Frei S, Dittmann MT, Furrer S, Osmann C, Ortmann S, et al. Digestive

physiology, metabolism and methane production of captive Linné’s two-toed sloths

(Choloepus didactylus). J Anim Physiol Anim Nutr (Berl). 2016;100(3):552–64.

8. Cerling T, Harris J. Carbon isotope fractionation betwwen diet and bioapatite in ungulate

mammals and implications for ecological and palaecological studies. Oecologia.

1999;120(3):347–63.

9. Martin JE, Vance D, Balter V. Magnesium stable isotope ecology using mammal tooth

enamel. Proc Natl Acad Sci [Internet]. 2015;112(2):430–5. Available from:

http://www.pnas.org/lookup/doi/10.1073/pnas.1417792112

10. Cerling TE, Andanje SA, Blumenthal SA, Brown FH, Chritz KL, Harris JM, et al. Dietary

changes of large herbivores in the Turkana Basin, Kenya from 4 to 1 Ma. Proc Natl Acad

Sci [Internet]. 2015;112(37):11467–72. Available from:

http://www.pnas.org/lookup/doi/10.1073/pnas.1513075112

11. Chikaraishi Y, Kashiyama Y, Ogawa NO, Kitazato H, Ohkouchi N. Metabolic control of

nitrogen isotope composition of amino acids in macroalgae and gastropods : implications

for aquatic food web studies. Mar Ecol Prog Ser. 2007;342:85–90.

12. Chikaraishi Y, Ogawa NO, Kashiyama Y, Takano Y, Suga H, Tomitani A, et al.

Determination of aquatic food-web structure based on compound-specific nitrogen

isotopic composition of amino acids. Limnol Oceanogr. 2009;(7):740–50.

13. Naito YI, Chikaraishi Y, Ohkouchi N, Dorothée G, Bocherens H, Naito YI, et al. Nitrogen

isotopic composition of collagen amino acids as an indicator of aquatic resource

consumption : insights from Mesolithic and Epipalaeolithic archaeological sites in France

Nitrogen isotopic composition of collagen amino acids as an indicator of aquatic resource

consumption : insights from Mesolithic and Epipalaeolithic archaeological sites in France.

2013;8243.

14. Naito YI, Chikaraishi Y, Drucker DoG, Ohkouchi N, Semal P, Wißing C, et al. Ecological

niche of Neanderthals from Spy Cave revealed by nitrogen isotopes of individual amino

acids in collagen. J Hum Evol. 2016;93:82–90.

15. O’Connell TCO. ‘Trophic’ and ‘source’ amino acids in trophic estimation: a likely

metabolic explanation. Oecologia. 2017;184(2):317–26.

16. Popp BN, Graham BS, Olson RJ, Hannides CCS, Lott MJ, Lopez-Ibarra GA, et al. Insight

into the trophic ecology of yellowfin tuna, Thunnus albacares, from compound- specific

nitrogen isotope analysis of proteinaceous amino acids. In: Dawson TE, Siegwolf RTW,

editors. Stable Isotopes as Indicators of Ecological Change. Elsevier Inc.; 2007. p. 173–

90.

17. Lee-Thorp JJ, Sealy JJC, van der Merwe NJN. Stable carbon isotope ratio differences

between bone collagen and bone apatite, and their relationship to diet. J Archaeol Sci.

1989;32:1459–70.

18. Clementz MT, Fox-Dobbs K, Wheatley P V, Koch PL, Doak DF. Revisiting old bones:

coupled carbon isotope analysis of bioapatite and collagen as an ecological and

palaeoecological tool. Geol J. 2009;44:605–20.

19. Bocherens H. Isotopic biogeochemistry and the paleoecology of the mammoth steppe

fauna. Deinsea. 2003;9(1):57–76.

20. McMahon KW, McCarthy MD. Embracing variability in amino acid d15N fractionation:

mechanisms, implications, and applications for trophic ecology. Ecosphere. 2016;7(12):1–

26.

21. Fariña RA. Trophic relationships among Lujanian mammals. Evol Theory.

1996;11(1839):125–34.

22. Fariña RA, Blanco RE. Megatherium the Stabber. Proc R Soc B Biol Sci. 2006;263:1725–

23. Bargo MS, De Iuliis G, Vizcaíno SF. Hypsodonty in Pleistocene ground sloths. Acta

Palaeontol Pol. 2006;51(1):53–61.

24. Vizcaino SF, Bargo MS, Cassini GH. Dental occlusal surface area in relation to body

mass, food habits and other biologic features in fossil xenarthrans. Ameghiniana.

2006;43(1):11–26.

25. Croft DA. The middle Miocene (Laventan) Quebrada Honda Fauna, southern Bolivia and

a description of its notoungulates. Palaeontology. 2007;50(1):277–303.

26. Lorrain A, Graham B, Ménard F, Popp B, Bouillon S, Breugel P Van, et al. Nitrogen and

carbon isotope values of individual amino acids : a tool to study foraging ecology of

penguins in the Southern Ocean. 2009;

27. Nielsen JM, Popp BN, Winder M. Meta ‑ analysis of amino acid stable nitrogen isotope

ratios for estimating trophic position in marine organisms. Oecologia [Internet].

2015;631–42. Available from: http://dx.doi.org/10.1007/s00442-015-3305-7

28. Chikaraishi Y, Ogawa NO, Ohkouchi N. Further evaluation of the trophic level estimation

based on nitrogen isotopic composition of amino acids. In: Ohkouchi N, Tayasu I, Koba

K, editors. Earth, Life, and Isotopes. Kyoto Universy Press; 2010. p. 37–51.

29. Cerling TE, Omondi P, Macharia AN. Diets of Kenyan elephants from stable isotopes and

the origin of confiscated ivory in Kenya. Afr J Ecol. 2007;45(4):614–23.

30. Sponheimer M, Robinson T, Ayliffe L, Passey B, Roeder B, Shipley L, et al. An

experimental study of carbon-isotope fractionation between diet, hair, and feces of

mammalian herbivores. Can J Zool [Internet]. 2003;81(5):871–6. Available from:

http://www.nrcresearchpress.com/doi/abs/10.1139/z03-066

31. Kohn MJ, Cerling TE. Stable Isotope Compositions of Biological Apatite. Rev Mineral

Geochemistry [Internet]. 2002;48(1):455–88. Available from:

http://rimg.geoscienceworld.org/cgi/doi/10.2138/rmg.2002.48.12

32. Darwin C. Journal of Researches into the Natural History and Geology of the Countries

Visited during the Voyage of the H.M.S Beagle Round the World. 2nd ed. New York:

Appleton; 1839. 556 p.

33. Krueger HW, Sullivan CH. Models for carbon isotope fractionation between diet and

bone. In: Turnlund JR, Johnson PE, editors. Stable Isotopes in Nutrition American

Chemical Society Symposium Series 258 Washington, DC: American Chemical Society.

1984. p. 205–20.

34. Sullivan CH, Krueger HW. Carbon isotope analysis of separate chemical phases in

modern and fossil bone. Vol. 292, Nature. 1981. p. 333–5.

35. Garvie-Lok SJ, Varney TL, Katzenberg MA. Preparation of bone carbonate for stable

isotope analysis: The effects of treatment time and acid concentration. J Archaeol Sci.

2004;31(8):763–76.

36. Lee-Thorp J, Sponheimer M. Three case studies used to reassess the reliability of fossil

bone and enamel isotope signals for paleodietary studies. J Anthropol Archaeol.

2003;22(3):208–16.

37. Snoeck C, Pellegrini M. Comparing bioapatite carbonate pre-treatments for isotopic

measurements: Part 1-Impact on structure and chemical composition. Chem Geol.

2015;417:349–403.

38. Pellegrini M, Snoeck C. Comparing bioapatite carbonate pre-treatments for isotopic

measurements: Part 2 - Impact on carbon and oxygen isotope compositions. Chem Geol.

2016;420:88–96.

39. Ambrose SH. Preparation and characterization of bone and tooth collagen for isotopic

analysis. J Archaeol Sci. 1990;17(4):431–51.

40. Phillips DL, Koch PL. Incorporating concentration dependence in stable isotope mixing

models. Oecologia [Internet]. 2002;130(1):114–25. Available from:

http://link.springer.com/10.1007/s004420100786

41. Kleiber M. Body size and metabolism. Vol. 6, Hilgardia: A Journal of Agricultural

Science. 1932. p. 315–53.

42. Gillooly JF, Brown JH, West GB, Savage VM, Charnov EL. Effects of size and

temperature on metabolic rate. Science (80- ). 2001;293:2248–51.

43. Rust F. Ruminant Methane Delta (δ13C / δ12C) Values: Relation to Atmospheric

Methane. Science (80- ). 1981;211(4486):1044–6.

44. Games LM, Gunsalus RP. Methane-producing bacteria : natural fractionations of the

stable carbon isotopes. Geochimica. 1978;42:1295–7.

45. Howard BH. Intestinal microorganisms of ruminants and other vertebrates. In: Henry SM,

editor. Symbiosis. New York: Academic Press; 1967. p. 317–67.

46. Parra R. Comparison of foregut and hindgut fermentation in herbivores. In: The Ecology

of arboreal folivores. Smitsonian Press; 1978. p. 205–29.

47. Xue Z, Zhang W, Wang L, Hou R, Zhang M, Fei L, et al. The bamboo-eating giant panda

harbors a carnivore-like gut microbiota, with excessive seasonal variations. MBio.

2015;6:e00022–15.

48. Hume ID. Gut morphology, body size and digestive performance in rodents. In: The

Digestive System in Mammals Food, Form and Function. Cambridge U Press; 1994. p.

315–23.

49. Franz R, Soliva CR, Kreuzer M, Hummel J, Clauss M. Methane output of rabbits

(Oryctolagus cuniculus) and guinea pigs (Cavia porcellus) fed a hay-only diet:

Implications for the scaling of methane production with body mass in non-ruminant

mammalian herbivores. Comp Biochem Physiol - A Mol Integr Physiol [Internet].

2011;158(1):177–81. Available from: http://dx.doi.org/10.1016/j.cbpa.2010.10.019

50. Eberhard IH. Ecology of the koala, Phascolarctos cinereus (Goldfuss) Marsupialia:

Phascolarctidae, in Australia. In: The Ecology of Arboreal Folivores. 1978. p. 315–28.

51. Montgomery GG. Habitat selection and use by two-toed and three-toed sloths. In: SThe

Ecology of arboreal folivores. 1978. p. 329–59.

52. Dill-McFarland KA, Weimer PJ, Pauli JN, Peery MZ, Suen G. Diet specialization selects

for an unusual and simplified gut microbiota in two- and three-toed sloths. Environ

Microbiol. 2016;18(5):1391–402.

53. Raines J. Captive Health and Husbandry of the Bradypodidae. Zoo Biol. 2005;24:557–68.

54. Pauli JN, Peery MZ, Fountain ED, Karasov WH. Arboreal Folivores Limit Their

Energetic Output, All the Way to Slothfulness. Am Nat [Internet]. 2016;188(2):196–204.

Available from: http://www.journals.uchicago.edu/doi/10.1086/687032

55. McNab BK. Energetics, population biology, and distribution of xenarthrans, living and

extinct. In: The ecology of arboreal folivores. Smithsonian Press; 1985. p. 219–32.

56. Goffart M. Function and form in the sloth: international series of monographs in pure and

applied biology. Pergamon Press; 1971.

57. Merritt DA. The two-toed Hoffman’s sloth, Choloepus hoffmanni Peters. In: The

evolution and ecology of armadillos, sloths, and vermilinguas. Smithsonian Press; 1985. p.

100

333–41.

58. Heymann EW, Flores Amasifuén C, Shahuano Tello N, Tirado Herrera ER, Stojan-Dolar

M. Disgusting appetite: two-toed sloths feeding in human latrines. Mamm Biol.

2011;76(1):84–6.

59. De Menezes AB, Lewis E, O’Donovan M, O’Neill BF, Clipson N, Doyle EM.

Microbiome analysis of dairy cows fed pasture or total mixed ration diets. FEMS

Microbiol Ecol. 2011;78(2):256–65.

60. Franz R, Soliva CR, Kreuzer M, Steuer P, Hummel J, Clauss M. Methane production in

relation to body mass of ruminants and equids. Evol Ecol Res. 2010;12:727–38.

61. Crutzen PJ, Aselmann I, Seiler W. Methane production by domestic animals, wild

ruminants, other herbivorous fauna, and humans. Tellus B. 1986;38 B(3–4):271–84.

62. Stevens CE, Hume ID. Contributions of microbes in vertebrate gastrointestinal tract to

production and conservation of nutrients. Physiol Rev. 1998;78(2):393–427.

63. McNab BK. The Energetics of Endotherms. Ohio J Sci [Internet]. 1974;74(6):370–80.

Available from: http://hdl.handle.net/1811/22251

64. Feer F. Observations ecologiques sur le neotrague de Bates (Neotragus batesi de Winton,

1903, Artiodactyle, Ruminant, Bovide) du nord-est du Gabon. Terre Vie Rev Ecol.

1979;33:159–239.

65. Kohn MJ, McKay MP. Paleoecology of late Pleistocene-Holocene faunas of eastern and

central Wyoming, USA, with implications for LGM climate models. Palaeogeogr

Palaeoclimatol Palaeoecol [Internet]. 2012;326–328:42–53. Available from:

http://dx.doi.org/10.1016/j.palaeo.2012.01.037

66. Zanazzi A, Judd E, Fletcher A, Bryant H, Kohn MJ. Eocene-Oligocene latitudinal climate

101

gradients in North America inferred from stable isotope ratios in perissodactyl tooth

enamel. Palaeogeogr Palaeoclimatol Palaeoecol [Internet]. 2015;417:561–8. Available

from: http://dx.doi.org/10.1016/j.palaeo.2014.10.024

67. Bargo MS, Toledo N, Vizcaino SF. Muzzle of South American Pleistocene ground sloths

(Xenarthra, Tardigrada). J Morphol. 2006;267:248–63.

68. Clauss M, Frey R, Kiefer B, Lechner-Doll M, Loehlein W, Polster C, et al. The maximum

attainable body size of herbivorous mammals: morphophysiological constraints on

foregut, and adaptations of hindgut fermenters. Oecologia. 2003;136(1):14–27.

69. Burnham RJ, Johnson KR. South American palaeobotany and the origins of neotropical

rainforests. Philos Trans R Soc B Biol Sci [Internet]. 2004;359(1450):1595–610.

Available from: http://rstb.royalsocietypublishing.org/cgi/doi/10.1098/rstb.2004.1531

70. Jaramillo C, Hoorn C, Silva SAF, Leite F, Herrera F, Quiroz L, et al. The origin of the

modern Amazon rainforest: implications of the palynological and paleobotanical record.

In: Hoorn C, Wesselingh FP, editors. Amazonia, Landscape and Species Evolution: A

Look into the Past. 1st ed. Blackwell; 2010. p. 317–34.

71. Martin JE, Vance D, Balter V. Magnesium stable isotope ecology using mammal tooth

enamel. Proc Natl Acad Sci [Internet]. 2015;112(2):430–5. Available from:

http://www.pnas.org/lookup/doi/10.1073/pnas.1417792112

72. Levin NE, Simpson SW, Quade J, Cerling TE, Frost SR. Herbivore enamel carbon

isotopic composition and the environmental context of Ardipithecus at Gona, Ethiopia. In:

Quade J, Wynn JG, editors. The Geology of Early Humans in the Horn of Africa:

Geological Society of America Special Paper 446. Geological Society of America; 2008.

p. 215–34.

102

73. Blakey RC. Gondwana paleogeography from assembly to breakup — A 500 m. y.

odyssey. In: Fielding CR, Frank TD, Isbell JL, editors. Resolving the Late Paleozoic Ice

Age in Time and Space: The Geological Society of America Special Paper. 2008. p. 1–28.

74. Granot R, Dyment J. The Cretaceous opening of the South Atlantic Ocean. Earth Planet

Sci Lett [Internet]. 2015;414:156–63. Available from:

http://dx.doi.org/10.1016/j.epsl.2015.01.015

75. Li ZX, Powell CM. An outline of the palaeogeographic evolution of the Australasian

region since the beginning of the Neoproterozoic. Earth Sci Rev. 2001;53:237–77.

76. Barnosky AD, Koch PL, Feranec RS, Wing SL, Shabel AB. Assessing the causes of Late

Pleistocene extinctions on the continents. Science (80- ) [Internet]. 2004;306:70–5.

Available from: d:%5CARTICLES%5Cdeja

graves%5CBarnosky_Koch_Feranec_Wing_Shabel_2004.pdf

77. Antoine PO, Salas-Gismondi R, Pujos F, Ganerød M, Marivaux L. Western Amazonia as

a hotspot of mammalian biodiversity throughout the Cenozoic. J Mamm Evol.

2017;24(1):5–17.

78. Guiraud R, Bosworth W, Thierry J, Delplanque A. Phanerozoic geological evolution of

Northern and Central Africa: An overview. J African Earth Sci. 2005;43:83–143.

79. Maglio VJ, Cooke HBS. Evolution of African mammals. Maglio VJ, Cooke HBS, editors.

1978. 641 p.

80. Martinelli LA, Almeida S, Brown IF, Moreira MZ, Victoria RL, Sternberg LSL, et al.

Stable carbon isotope ratio of tree leaves, boles and fine litter in a tropical forest in

Rondonia, Brazil. Oecologia. 1998;(114):170–9.

81. Mortillaro JM, Pouilly M, Wach M, Freitas CEC, Abril G, Meziane T. Trophic

103

opportunism of central Amazon floodplain fish. Freshw Biol. 2015;60(8):1659–70.

82. Medina E, Minchin P. Stratification of δ 13C values of leaves in Amazonian rainforests.

Oecologia. 1980;45(3):377–8.

83. Buchmann N, Guehl JM, Barigah TS, Ehleringer JR. Interseasonal comparison of CO2

concentrations, isotopic composition, and carbon dynamics in an Amazonian rainforest

(French Guiana). Oecologia. 1997;110(1):120–31.

84. Oliveira ACB, Soares MGM, Martinelli LA, Moreira MZ. Carbon sources of fish in an

Amazonian floodplain lake. Aquat Sci. 2006;68(2):229–38.

85. van der Merwe NJ, Medina E. The canopy effect, carbon isotope ratios and foodwebs in

amazonia. J Archaeol Sci. 1991;18(3):249–59.

86. Graham H V., Patzkowsky ME, Wing SL, Parker GG, Fogel ML, Freeman KH. Isotopic

characteristics of canopies in simulated leaf assemblages. Geochim Cosmochim Acta.

2014;144:82–95.

87. Medina E, Martinelli LA, Barbosa E, Victoria RL. Natural abundance of 13C in tropical

grasses from the INPA, Instituto Nacional de Pesquisas da Amazônia, herbarium. Rev

Bras Botânica. 1999;22(1):44–51.

88. DeSantis LG. Stable isotope ecology of extant tapirs from the Americas. Biotropica.

2011;43(6):746–54.

89. Benedito-Cecilio E, Araujo-Lima CARM, Forsberg BR, Bittencourt MM, Martinelli LC.

Carbon sources of Amazonian fisheries. Fish Manag Ecol. 2000;7(4):305–15.

90. Magnusson WE, Araujo M, Cintra R, Lima A, Martinelli L, Sanaiotti T, et al.

Contributions of C3 and C4 plants to higher trophic levels in an Amazonian savanna.

1999;91–6.

104

91. Fair JM, Ryder TB, Loiselle BA, Blake JG, Larson toti E, Davis P, et al. Estimates of

dietary overlap for six species of Amazonian manakin birds using stable isotopes. Isotopes

Environ Health Stud. 2013;49(3):420–35.

92. Tejada-Lara J V., MacFadden BJ, Bermudez L, Rojas G, Salas-Gismondi R, Flynn JJ.

Body mass predicts isotope enrichment in herbivorous mammals. Proc R Soc B Biol Sci.

2018;285(1881).

93. Clementz MT, Koch PL, Beck CA. Diet induced differences in carbon isotope

fractionation between sirenians and terrestrial ungulates. Mar Biol. 2007;151(5):1773–84.

94. Ehleringer JR, Field CB, Lin Z, Kuo C, Ehleringer AJR, Field CB, et al. Leaf carbon

isotope and mineral composition in subtropical plants along an irradiance cline.

Oecologia. 1986;70(4):520–6.

95. Farquhar GD, Ehleringer JR, Hubick KT. Carbon isotope discrimination and

photosynthesis. Annu Rev Plant Physiol Plant Mol Biol. 1989;40:503–37.

96. Cerling TE, Harris JM, Macfadden BJ, Leakey MG, Quadek J, Eisenmann V, et al. Global

vegetation change through the Miocene / Pliocene boundary. Nature. 1997;389:153–8.

97. Cabral MMM, Zuanon J, Mattos GE De, Rosas FCW. Feeding habits of giant otters

Pteronura brasiliensis (Carnivora: Mustelidae) in the Balbina hydroelectric reservoir,

Central Brazilian Amazon. Zoologia. 2010;27(1):47–53.

98. Kohn MJ. Carbon isotope compositions of terrestrial C3 plants as indicators of

(paleo)ecology and (paleo)climate. Proc Natl Acad Sci [Internet]. 2010;107(46):19691–5.

Available from: http://www.pnas.org/cgi/doi/10.1073/pnas.1004933107

99. Schoeninger MJ, DeNiro MJ. Nitrogen and carbon isotopic composition of bone collagen

from marine and terrestrial animals. Geochim Cosmochim Acta. 1984;48:625–39.

105

100. Hedges REM. On bone collagen-apatite carbonate isotopic relationships. Int J

Osteoarchaeol. 2003;13(1–2):66–79.

101. O’Connell TC, Hedges REM. Chicken and egg: testing the carbon isotopic effects of

carnivory and herbivory. Archaeometry. 2017;59(2):302–15.

102. Shipley LA, Forbey JS, Moore BD. Revisiting the dietary niche: When is a mammalian

herbivore a specialist? Integr Comp Biol. 2009;49(3):274–90.

103. Berto A, da Silva AF, Visentainer J V, Matsushita M, DaSouza NE, Souza D. Proximate

compositions, mineral contents and fatty acid compositions of native Amazonian fruits.

Food Res Int [Internet]. 2015;77:441–9. Available from:

http://dx.doi.org/10.1016/j.foodres.2015.08.018

104. Nie Y, Wei F, Zhou W, Hu Y, Senior AM, Wu Q, et al. Giant pandas are macronutritional

carnivores. Curr Biol. 2019;29:1–6.

105. Codron D, Clauss M, Codron J, Tütken T. Within trophic level shifts in collagen –

carbonate stable carbon isotope spacing are propagated by diet and digestive physiology in

large mammal herbivores. Ecol Evol. 2018;8:3983–95.

106. Kellner CM, Schoeninger MJ. A simple carbon isotope model for reconstructing

prehistoric human diet. Am J Phys Anthropol. 2007;133:1112–27.

107. Machovsky-Capuska GE, Senior AM, Simpson SJ, Raubenheimer D. The

multidimensional nutritional niche. Trends Ecol Evol. 2016;31(5):355–65.

108. Defler TR. A history of terrrestrial mammals in South America. Topics in. Landman NH,

Harries PJ, editors. Cham, switzerland: Springer; 2019. 372 p.

109. Voss RS, Emmons LH. Mammalian diversity in Neotropical lowland rainforests : a

preliminary assessment. Bull Am Museum Nat Hist [Internet]. 1996;230:1–115. Available

106

from: http://orton.catie.ac.cr/cgi-

bin/wxis.exe/?IsisScript=OET.xis&method=post&formato=2&cantidad=1

&expresion=mfn=013548

110. Holt RD. Bringing the Hutchinsonian niche into the 21st century: ecological and

evolutionary perspectives. Proc Natl Acad Sci. 2009;106:19659–65.

111. Marin J, Rapacciuolo G, Costa GC, Graham CH, Brooks TM, Young BE, et al.

Evolutionary time drives tetrapod diversity. Proc R Soc B. 2018;20172378.

112. Barnosky AD, Lindsey EL, Villavicencio NA, Bostelmann E, Hadly EA, Wanket J, et al.

Variable impact of late-Quaternary megafaunal extinction in causing ecological state

shifts in North and South America. Proc Natl Acad Sci U S A. 2016;113(4):856–61.

113. MacFadden BJ. Diet and habitat of toxodont megaherbivores (Mammalia, Notoungulata)

from the late Quaternary of South and Central America. Quat Res. 2005;64(2):113–24.

114. Bocherens H, Cotte M, Bonini RA, Straccia P, Scian D, Soibelzon L, et al. Isotopic insight

on paleodiet of extinct Pleistocene megafaunal Xenarthrans from Argentina. Gondwana

Res [Internet]. 2017;48(April):7–14. Available from:

http://dx.doi.org/10.1016/j.gr.2017.04.003

115. Domingo L, Tomassini RL, Montalvo CI, Sanz-Pérez D, Alberdi MT. The Great

American Biotic Interchange revisited: a new perspective from the stable isotope record of

Argentine Pampas fossil mammals. Sci Rep. 2020;10(1608):1–10.

116. Kohn MJ, McKay MP, Knight JL. Dining in the Pleistocene - Who’s on the menu?

Geology. 2005;33(8):649–52.

117. Croft DA, Engelman RK, Dolgushina T, Wesley G. Diversity and disparity of

sparassodonts (Metatheria ) reveal non-analogue nature of ancient South American

107

mammalian carnivore guilds. Proc R Soc B Biol Sci. 2018;285:20172012.

118. de Muizon C, McDonald HG. An aquatic sloth from the Pliocene of Peru. Vol. 375,

Nature. 1995. p. 224–7.

119. Boecklen WJ, Yarnes CT, Cook BA, James AC. On the use of stable isotopes in trophic

ecology. Annu Rev Ecol Evol Syst. 2011;42:411–40.

120. McClelland JW, Montoya JP. Trophic relationships and the nitrogen isotopic composition

of amino acids in plankton. Ecology. 2002;83(8):2173–80.

121. Chikaraishi Y, Ogawa NO, Doi H, Ohkouchi N. 15N/14 N ratios of amino acids as a tool

for studying terrestrial food webs: a case study of terrestrial insects (bees, wasps, and

hornets ). Ecol Res. 2011;26:835–44.

122. Nielsen JM, Popp BN, Winder M. Meta‑analysis of amino acid stable nitrogen isotope

ratios for estimating trophic position in marine organisms. Oecologia [Internet].

2015;631–42. Available from: http://dx.doi.org/10.1007/s00442-015-3305-7

123. Kendall IP, Lee MRF, Evershed RP. The effect of trophic level on individual amino acid

δ15N values in a terrestrial ruminant food web. Sci Technol Archaeol Res.

2017;3(1):135–45.

124. Steffan SA, Chikaraishi Y, Horton DR, Ohkouchi N, Singleton ME, Miliczky E, et al.

Trophic Hierarchies Illuminated via Amino Acid Isotopic Analysis. 2013;8(9):21–3.

125. Metabolism C, Brosnan JT. Glutamate and Glutamine in Metabolism Glutamate , at the

Interface between Amino Acid and. 2000;(1992):988–90.

126. Styring AK, Fraser RA, Bogaard A, Evershed RP. Cereal grain, rachis and pulse seed

amino acid d15N values as indicators of plant nitrogen metabolism. Phytochemistry

[Internet]. 2014;97:20–9. Available from:

108

http://dx.doi.org/10.1016/j.phytochem.2013.05.009

127. Paolini M, Ziller L, Laursen KH, Camin F. Compound-Speci fi c δ 15 N and δ 13 C

Analyses of Amino Acids for Potential Discrimination between Organically and

Conventionally Grown Wheat. 2015;

128. Steffan SA, Chikaraishi Y, Currie CR, Horn H, Gaines-day HR, Pauli JN, et al. Microbes

are trophic analogs of animals. 2015;112(49):15119–24.

129. Bloomfield AL, Elsdon TS, Walther BD, Gier EJ, Gillanders BM. Temperature and diet

affect carbon and nitrogen isotopes of fish muscle: can amino acid nitrogen isotopes

explain effects? J Exp Mar Bio Ecol. 2011;399:48–59.

130. Hoen DK, Kim SL, Hussey NE, Wallsgrove NJ, Drazen JC, Popp BN. Amino acid 15N

trophic enrichment factors of four large carnivorous fishes. J Exp Mar Bio Ecol [Internet].

2014;453:76–83. Available from: http://dx.doi.org/10.1016/j.jembe.2014.01.006

131. Dale JJ, Wallsgrove NJ, Popp BN, Holland KN. Nursery habitat use and foraging ecology

of the brown stingray Dasyatis lata determined from stomach contents, bulk and amino

acid stable isotopes. Mar Ecol Prog Ser. 2011;433:221–36.

132. Germain LR, Koch PL, Harvey J, Mccarthy MD. Nitrogen isotope fractionation in amino

acids from harbor seals: implications for compound-specific trophic position calculations.

Mar Ecol Prog Ser. 2013;482:265–77.

133. Connell TCO, Collins MJ. Comment on “ Ecological niche of Neanderthals from Spy

Cave revealed by nitrogen isotopes of individual amino acids in collagen .” J Hum Evol

[Internet]. 2017;93:2016–8. Available from:

http://dx.doi.org/10.1016/j.jhevol.2017.05.006

134. Owen-Smith RN. Megaherbivores: the influence of very large body size on ecology.

109

Cambridge U Press; 1988. 369 p.

135. Janis C. An estimation of tooth volume and hypsodonty indices in ungulate mammals, and

the correlation of these factors with dietary preferences. In: Russell DE, Santoro JP,

Sigogneau-Russell D, editors. Teeth revisited: Proceedings of the VIIth International

Symposium on dental morphology. Paris: Mem. Mus. Nat. Hist.; 1988. p. 367–87.

136. Janis CM. Correlations between craniodental anatomy and feeding in ungulates: reciprocal

illumination between living and fossil taxa. Funct Morphol Vertebr Paleontol. 1995;76–

98.

137. Naples VL. Reconstruction of cranial morphology and analysis of function in the

Pleistocene ground sloth Nothrotheriops shastensis (Mammalia, Megatheriidae). Contrib

Sci. 1987;389:1–21.

138. Bargo MS, Vizcaíno SF. Paleobiology of Pleistocene ground sloths (Xenarthra ,

Tardigrada): biomechanics, morphogeometry and ecomorphology applied to the

masticatory apparatus. Ameghiniana. 2008;45(1):175–96.

110

Appendix A

Supplementary information for Chapter 1

Stable isotope analyses

Samples of teeth and feces of Bradypus variegatus and Choloepus hoffmanni were collected from adult individuals raised at the Huachipa Zoo (Lima, Peru), except for one dental sample of

Choloepus hoffmanni from the Tampa’s Lowry Park Zoo (Florida, USA). Individuals were either born at or adopted by the zoos at a young age, and spent several years being fed a controlled diet

(spanning 5 to 12 years under a controlled diet feeding regime at the zoo). Samples of living specimens from the Huachipa Zoo were collected by the zoo staff during the annual health inspection. Dental bioapatite was analyzed for d13C. Choloepus and Bradypus have ever-growing teeth which in captivity are not sufficiently worn down over time as they are in natural conditions. During health inspections at zoos, excessively grown caniniforms (Choloepus) and incisiviforms (Bradypus) are trimmed to prevent animals from hurting themselves (in a similar fashion to veterinarian treatments for hamsters). Instead of being discarded, these disposable samples of dental tissues were collected and used for isotopic analyses. Samples were later powdered in the lab using a mortar and pestle.

Powdered samples of dental bioapatite were analyzed for d13C and d18O. Modern samples were first treated for three hours with H2O2 to remove organic contaminants, rinsed three times with distilled water, bathed for an hour in 0.1 N acetic acid to remove any secondary carbonates, rinsed three more times with distilled water followed by a fourth rinse with methanol, and finally dried. Samples were run on a Finnigan MAT 252 isotope ratio mass spectometer (IRMS) with a

Kiel device. Foodstuff and feces were collected on a daily basis for 11 and 15 days respectively,

111 and analyzed for d13C and d15N. All organic samples were freeze dried and homogenized in a

Spex 67000 liquid nitrogen mill. Weight %C and %N were determined using a Carlo-Erba 1500 elemental analyzer after IRMS analysis. All the samples were processed and analyzed at the

Department of Geological Sciences, University of Florida (Gainesville).

Fossil dung was freeze dried, homogenized, and processed as for the modern organic samples. Teeth of fossil sloths were sampled using a Dremel® drill and carbide dental burrs.

Approximately 0.1 g of powdered outermost dentine was collected and processed as described above.

Values are reported using the conventional permil (‰) notation wherein δ13C =

13 12 13 (Rsample/Rstandard – 1) x 1000), and R= C/ C. Values of δ C are reported relative to the VPDB standard. Isotope enrichment (Ɛ*) was calculated as in Cerling and Harris [3], i.e., as a function of α* (apparent fractionation), as follows: Ɛ* = (α* - 1) x 1000, where α* = (1000 +

13 13 δ Ctissue)/(1000 + δ Cdiet). The superscript (*) indicates that isotopic equilibrium is not assumed.

Analytical precision of standards:

Inorganic samples (teeth bioapatite, NBS-19 standard) d13C = 0.02; d18O = 0.05

Organic samples (food and feces; USGS-40 standard): d13C = 0.09; d15N = 0.22

112

Foodstuff

A concentration-weighted linear mixing model [49] was used to calculate the isotopic signatures of any mixed diet. In this model, the proportional contribution of each component to the mixture (and therefore to the consumer) equals the fraction of that component in the mixture multiplied by its elemental concentration (in this case, wt %C and wt %N) and divided by the sum of the products of fraction and concentration of each dietary component, as follows:

�,[�] �, = �,[�] + �,[�] + �,[�]

Where �(x, C); �(y, C), �(z, C), are the fractions of assimilated C, of components “x”, “y”,

“z” in the mixed diet.

Regression models and traitgram

Data for linear regression models of mammals other than sloths, were collected from the literature (table 2, table S1). Values of e*diet-bioapatite were regressed against body mass (BM), basal rate of metabolism (BMR), average rectal temperature, and range of rectal temperature variation. All data were log transformed (ln). Statistics and regression modeling were performed with RStudio (v 1.0.136) using packages MASS, foreign, quantreg, MuMIn, and lme4. Traitgram

113 was performed with the R package phytools [50]. We assessed relationships between variables by performing OLS, quantile, and robust regression models; Cook’s distance was used as an estimate for the influence of outliers. Linear mixed effect model analyses based on AICc tests were used to assess whether a single variable, reduced model, or various combinatorial model subsets were better supported by the data than a complete global model that includes all the potential predictors. Tree topology for Figure 1 is from O’Leary et al. [51]. Divergence ages (in

Ma, or Megannum), only for illustrative purposes (providing approximations for visual representation of the pattern of lineage splitting, but not used in calculations), are mean estimates from the literature (see below).

Regression formulae

This paper proposes three regression formulae (body-mass calibrated) to calculate the C diet-bioapatite enrichment (e*diet-bioapatite) of herbivorous mammals (both extant and extinct). The first formula (A) is based on the regression analysis of correlation of all mammals included:

e ∗ = �. � + �. �� (��) (A)

Where BM is the body mass in kg and should be log transformed (ln). The obtained diet-

x bioapatite e* will need to be inverted (e ) to obtain the ‰ value to be applied for interpretation of the isotopic signature of the herbivorous mammal under study.

114

This formula (A) should be used when the type of digestive fermentation of the mammal under study is unknown, does not fall within foregut or hindgut types of fermentation (e.g. giant panda), or to get a general or most conservative body-size calibrated value for e*diet-bioapatite.

The two other formulae we propose separate mammals by foregut (B) versus hindgut fermenters

(C):

e ∗ = �. �� + �. �� (��) (B)

e ∗ = �. �� + �. �� (��) (C)

These formulae should be applied when the type of digestive fermentation (foregut vs hindgut) is known or most likely given phylogenetic history of the mammal under study. As for the first

x formula, BM is in kg and log transformed (ln), and e* needs to be inverted (e ) to obtain the ‰ value to be applied for interpretation of the isotopic signature of the herbivore mammal under study. Table 3 documents that multiple models agree closely in terms of regression slopes and intercepts. Rather than select one arbitrarily, we advocate instead for applying a model averaging approach. As discussed by Sears et al. [52], model averaging is appropriate when there is no single, clearly identifiable “correct” model, and given the close agreement across all of the models and that each performs equally well in estimating the variables, use of the mean of the individual estimates is appropriate here.

115

Sources of divergence ages

Bradypus / Choloepus+Mylodon: 27 ± 3 Ma [53]; or 29.3 Ma sensu [54].

Mylodon / Choloepus: 27.5 Ma [54].

Muridae (Mus) / Cricetidae (Microtus): 65.8 Ma [55].

Rabbit / Mouse: 65.5 Ma; based on Mimotona wana from [51]

Giraffidae / remaining Pecora (i.e. including Bos): 19 Ma [56].

Suidae / Ruminantia: 35.2 Ma; based on Elomeryx crispus from [51].

Ruminantia / Camelidae: 49-55 Ma; (based on Pseudamphimeryx, first ruminant, [57] from

[58]).

Camelus / Lama: 25 Ma [59].

Equus / Diceros: 52-58 Ma; split Ceratomorpha and Hippomorpha [60].

Equus caballus / Equus burchelli: 3 Ma [61].

Arytiodactyla / Perissodactyla: 55.4-50.3 Ma, based on Hyracotherium angustidens [51].

Carnivora / Eungulata: 84.9 Ma [62].

Afrotheria: 101 Ma [62].

Euarchontoglires / Laurasiatheria: 98.9 Ma [62].

Xenarthra / Epitheria: 101 Ma [62].

Marsupialia / Eutheria: 147.7 Ma [62].

Sources of life history trait values

BM: body mass; BMR: basal metabolic rate, and basal metabolic rate corrected for body mass using the Kleiber value; Temp: body temperature, as average rectal temperature and breadth of

116 range of rectal temperature. † = extinct taxon. Unpublished data from [6] provided by the lead author of that study, B. Passey.

Choloepus hoffmanni

BM: data from this study; BMR [35]; Temp: data from this study and [36].

Notes: Range of temperature is from [36] but average temperature is from specimens analyzed in this study (Huachipa Zoo).

Bradypus variegatus

BM: data from this study; BMR [35], Temp: data from this study and [36].

Notes: same as for Choloepus.

†Mylodon darwinii

BM [63].

Note: Although McNab (1985, [35]) provided estimates for BMR and BMR/Kleiber value for

Mylodon listai, we consider these estimations as unreliable because they are based on highly questionable foundations (body hair length as a proxy for thermal insulation correlated to BMR).

Bos taurus

BM [unpublished data from 6]; BMR [64]; Temp [65].

Sus scrofa

BM [unpublished data from 6]; BMR [64]; Temp [66].

117

Oryctolagus cuniculus

BM [unpublished data from 6]; BMR [67]; Temp [68].

Notes: BMR based on a close relative, Lepus alleni.

Microtus ochrogaster

BM [unpublished data from 6]; BMR [69]; Temp [69,70].

Notes: BMR based on a congener, Microtus guentheri.

Mus musculus

BM [5]; BMR [64]; Temp [71,72]

Lama guanicoe

BM [73]; BMR [73,74]; Temp [75].

Notes: BMR based on a congener, Lama glama.

Giraffa camelopardalis

BM [76]; BMR [77]; Temp [78].

Notes: BMR calculated from resting metabolic rate (RMR). BMR was assumed to account for

~85% of the RMR.

Diceros bicornis

BM [76,79]; Temp [80].

118

Camelus bactrianus

BM [67,76,79]; BMR [67]; Temp [81].

Notes: BMR based on a congener, Camelus dromedarius.

Equus caballus

BM [82]; BMR [64]; Temp [83].

Equus burchelli

BM [84]; BMR [64]; Temp [84].

Notes: BMR based on a congener, E. caballus.

Ailuropoda melanoleuca

BM [85]; BMR [85]; Temp [86].

Notes: BMR calculated from resting metabolic rate.

Loxodonta africana

BM [82]; BMR [64]; Temp [87].

Notes: BMR based on a close relative, Elephas maximus

Phascolarctos cinereus

BM [67,82]; BMR [67]; Temp [88].

119

BM FER BMR/Klei Temp Temp Temp TAXA E*(mean) BMR (Kg) M ber value Mean Variation Range

Choloepus hoffmanni 12.6 8.28 fore 0.188 0.44 34.4 33.4-36.2 2.8

Bradypus variegatus 10.3 4.23 fore 0.181 0.42 31 28.4-37.6 9.2

Mylodon darwinii† 15.6 1600 fore NA NA NA NA NA

Bos taurus 14.6 322.7 fore 0.17 1.14 38.7 38.3-39.1 0.8

Sus scrofa 13.3 128.6 hind 0.11 0.53 39.3 38.8-39.8 1

Oryctolagus cuniculus 12.8 3.5 hind 0.45 1 39.3 38.1-40.8 2.7

Microtus ochrogaster 11.5 0.05 hind 1.18 0.92 38 37.3-38.7 1.4

Mus musculus 9.1 0.021 hind 3.4 2.12 37.65 36-39.3 3.3

Lama guanicoe 12.9 120 fore 0.23 1.24 38 37.5-38.5 1

Giraffa camelopardalis 14.1 1500 fore 0.13 1.08 38.5 38-39 1

Diceros bicornis 14.4 1089 hind NA NA 37.74 36.8-38.6 1.8

Camelus bactrianus 13.7 454 fore 0.1 0.74 37 34-40 6

Equus caballus 13.7 260 hind 0.25 1.65 37.7 37.2-38.2 1

Ailuropoda melanoleuca 10.1 92 NA 0.082 0.42 37.3 37-37.6 0.6

Phascolartos cinereus 10.3 4.8 hind 0.22 0.54 36 34.2-37.7 3.5

Loxodonta africana 14.3 3600 hind 0.15 1.78 36.5 36-37 1

Equus burchelli 13.2 280 hind NA NA 39.3 38.4-41.8 3.4

Table S1. Summary of the e*diet-bioapatite values (highlighted box) for taxa in this study and other life history trait values. Taxonomic selection was driven by: (1) accuracy of dietary d13C data and (2) sampling of phylogenetic diversity. Abbreviations: BM: body mass (kg), FERM: type of 3 digestive system (foregut vs hindgut fermentation), BMR: basal rate of metabolism (cm O2/g*h), Temp: body temperature (°C). Sources for all values above and in Table S2. † is a recently extinct Pleistocene sloth.

120

TAXA e* Controlled Source of e* Choloepus hoffmanni 12.62 ± 0.68 Yes this study Bradypus variegatus 10.31 ± 1.03 Yes this study Mylodon darwinii† 15.63± 0.51 Yes, (diet from dung) this study Bos taurus 14.6± 0.3 Yes [6] Sus scrofa 12.9± 0.5 Yes [6] Oryctolagus cuniculus 12.8± 0.7 Yes [6] Microtus ochrogaster 11.5± 0.3 Yes [6] Mus musculus 9.1± 1.6 Yes [5] Lama guanicoe 12.9 No [3] Giraffa camelopardalis 14.1 No [3] Diceros bicornis 14.4 No [3] Camelus bactrianus 13.7 Partial [3] Equus caballus 13.7 No [3] Loxodonta africana 14.3 No [3] Ailuropoda melanoleuca 10.1 Partial [89] Equus burchelli 13.2 No [90] Phascolartos cinereus 10.3 Partial [91]

Table S2. Summary of the e*diet-bioapatite values included in this study. Selection of taxa was determined by: (1) dietary d13C controlled, (2) partially controlled (i.e., monospecific diets), or

13 (3) not controlled but still with a narrow range of reported d Cdiet variation in the wild. † is an extinct taxon.

121

AICc Brownian Motion -11.17615193 Lambda -9.064478475 Delta -10.73648238 Kappa -8.28545515 Early Burst -8.186953844 Ornstein Uhlenbeck -10.79611806 White Noise -11.93408278

Table S3. Summary of AIC analyses comparing different models of trait evolution. Lowest AICc score (gray highlight) corresponds to a White Noise model, which disregards phylogeny. The second lowest AICc score corresponds to Brownian motion (BM), which further emphasizes the lack of phylogenetic signal in the known values of e*diet-bioapatite across mammals. Data were fitted to the framework phylogenetic tree with the R package GEIGER [92]. Refer to [92] for explanation of each model of trait evolution.

e* vs: Body Mass BMR BMR/Kleiber Temperature Breadth of value (average) temp variation R2 p- R2 p- R2 p- R2 p- R2 p- value value value value value All taxa 0.62 0.000 0.24 0.078 0.06 0.366 0.16 0.122 0.12 0.19 Foregut 0.78 0.008 0.15 0.454 0.56 0.089 0.89 0.005 0.51 0.11 Hindgut 0.74 0.003 0.56 0.054 0.00 0.975 0.07 0.482 0.36 0.085

Table S4. Summary statistics for all regression analyses. Highlighted boxes indicate significant values. Only body mass correlated significantly with e* in all cases (i.e., in analyses with all taxa included as well as for mammals separated by type of digestive system [foregut and hindgut only]). Similar results (no significant correlations of e* with BMR corrected for body mass, for all mammals or either digestive system subset) when applying the alternative body mass scaling function of [93]; Kleiber value remains the most widely accepted scaling function for BMR/BM covariance, and we present it here because [93] excludes all ruminants, which are a substantial number of the taxa in our analyses and prior studies of carbon isotope enrichment in dental bioapatite. 122

Figure S1. Correlation between body mass (BM) and average rectal temperature. Values are log transformed. All mammals (black dashed line): R2=0.04, p-value=0.47; hindgut fermenters (red line and symbols): R2=0.00, p-value=0.95, foregut fermenters (greenish line and symbols): R2= 0.80, p-value= 0.02.

123

Supplementary References:

50. Revell LJ. 2011 phytools: an R package for phylogenetic comparative biology (and other things). Methods Ecol Evol 3, 217–223. (doi:10.1111/j.2041-210x.2011.00169.x) 51. O’Leary MA et al. 2013 The Placental mammal ancestor and the post-K-Pg radiation of placentals. Science 339, 662–667. (doi:10.1126/science.1229237) 52. Sears KE, Finarelli JA, Flynn JJ, Wyss AR. 2008 Estimating body mass in New World “monkeys” (Platyrrhini, Primates), with a consideration of the Miocene platyrrhine, Chilecebus carrascoensis. American Museum Novitates 3617:1–29. (doi:10.1206/627.1) 53. Delsuc F, Superina M, Tilak M-K, Douzery EJP, Hassanin A. 2012 Molecular phylogenetics unveils the ancient evolutionary origins of the enigmatic fairy armadillos. Molecular Phylogenetics and Evolution 62, 673–680. (doi:10.1016/j.ympev.2011.11.008) 54. Slater GJ, Cui P, Forasiepi AM, Lenz D, Tsangaras K, Voirin B, de Moraes-Barros N, MacPhee RDE, Greenwood AD. 2016 Evolutionary relationships among extinct and extant sloths: the evidence of mitogenomes and retroviruses. Genome Biology and Evolution 8, 607–621. (doi:10.1093/gbe/evw023) 55. Hedges SB, Kumar S. 1998 A molecular timescale for vertebrate evolution. Nature 392, 917–920. (doi:10.1038/31927) 56. Bibi F. 2013 A multi-calibrated mitochondrial phylogeny of extant Bovidae (Artiodactyla, Ruminantia) and the importance of the fossil record to systematics. BMC Evolutionary Biology 13, 166. (doi:10.1186/1471-2148-13-166) 57. Savage DE, Russell DE. 1983 Mammalian paleofaunas of the world. Addison-Wesley, London. 58. Theodor JM. 2004 Clock divergence estimates and the fossil record of Cetartiodactyla. J. Paleo. 78, 39–44. (10.1666/0022-3360(2004)078<0039:MCDEAT>2.0.CO;2) 59. Cui P, Ji R, Ding F, Qi D, Gao H, Meng H, Yu J, Hu S, Zhang H. 2007 A complete mitochondrial genome sequence of the wild two-humped camel (Camelus bactrianus ferus): an evolutionary history of camelidae. BMC Genomics 8, 241. (doi:10.1186/1471-2164-8- 241) 60. Steiner CC, Ryder OA. 2011 Molecular phylogeny and evolution of the Perissodactyla. Zoological Journal of the Linnean Society 163, 1289–1303. (doi:10.1111/j.1096- 3642.2011.00752.x) 61. Ishida N, Oyunsuren T, Mashima S, Mukoyama H, Saitou N. 1995 Mitochondrial DNA sequences of various species of the genus Equus with special reference to the phylogenetic relationship between Przewalskii’s wild horse and domestic horse. Journal of Molecular Evolution 41, 180–188. (doi:10.1007/bf00170671) 62. Bininda-Emonds ORP et al. 2007 The delayed rise of present-day mammals. Nature 446, 507–512. (doi:10.1038/nature05634) 63. Vizcaíno SF, Bargo MS, Cassini GH. 2006 Dental occlusal surface area in relation to body mass, food habits and other biological features in fossil xenarthrans. Ameghiniana 43,11–26.

124

64. Eisenberg JF. 1981 The mammalian radiation: An analysis of trends in evolution, adaptation, and behavior. U. Chicago Press, Chicago. 65. Espinoza JL, Sanchez J, Gracia JA, Sanchez JR, Ortega R, Palacios A. 2009 Thermoregulation differs in Chinampo (Bos taurus) and locally born dairy cattle. Turkish Journal of Veterinary and Animal Sciences 33, 175–180. 66. Jara AL, Hanson JM, Gabbard JD, Johnson SK, Register ET, He B, Tompkins SM. 2016 Comparison of Microchip Transponder and Noncontact Infrared Thermometry with Rectal Thermometry in domestic swine (Sus scrofa domestica). Journal of the American Association for Laboratory Animal Science 55, 588–593. 67. McNab BK. 1986 The Influence of Food Habits on the Energetics of Eutherian Mammals. Ecological Monographs 56, 1–19. (doi:10.2307/2937268) 68. Chen PH, White CE. 2006 Comparison of rectal, microchip transponder, and infrared thermometry techniques for obtaining body temperature in the laboratory rabbit (Oryctolagus cuniculus). Journal of the American Association for Laboratory Animal Science 45, 57–63. 69. Banin D, Haim A, Arad Z. 1994 Metabolism and thermoregulation in the Levant vole Microtus guentheri: The role of photoperiodicity. Journal of Thermal Biology 19, 55–62. (doi:10.1016/0306-4565(94)90009-4) 70. Cherry RH, Verner L. 1975 Seasonal Acclimatization to Temperature in the Prairie Vole, Microtus ochrogaster. American Midland Naturalist 94, 354. (doi:10.2307/2424431) 71. Hudson JW, Scott IM. 1979 Daily Torpor in the Laboratory Mouse, Mus musculus var. Albino. Physiological Zoology 52, 205–218. (doi:10.1086/physzool.52.2.30152564) 72. Rhodes JS, Koteja P, Swallow JG, Carter PA, Garland T Jr. 2000 Body temperatures of house mice artificially selected for high voluntary wheel-running behavior: repeatability and effect of genetic selection. Journal of Thermal Biology 25, 391–400. (doi:10.1016/s0306- 4565(99)00112-6) 73. Hayssen V, Lacy RC. 1985 Basal metabolic rates in mammals: Taxonomic differences in the allometry of BMR and body mass. Comparative Biochemistry and Physiology Part A: Physiology 81, 741–754. (doi:10.1016/0300-9629(85)90904-1) 74. El-Nouty F., Yousef M., Magdub A., Johnson H. 1978 Thyroid hormones and metabolic rate in burros, Equus asinus, and llamas, Lama glama: Effects of environmental temperature. Comparative Biochemistry and Physiology Part A: Physiology 60, 235–237. (doi:10.1016/0300-9629(78)90238-4) 75. de Lamo DA, Lacolla D, Heath JE. 2001 Sweating in the guanaco (Lama guanicoe). Journal of Thermal Biology 26, 77–83. (doi:10.1016/s0306-4565(00)00014-0) 76. Nowak RM. 1999 Walker's Mammals of the World, vol. 1. JHU Press. 77. Langman VA, Bamford OS, Maloiy GMO. 1982 Respiration and metabolism in the giraffe. Respiration Physiology 50, 141–152. (doi:10.1016/0034-5687(82)90013-5) 78. Mitchell G, Skinner J. 2004 Giraffe thermoregulation: a review. Transactions of the Royal Society of South Africa 59, 109–118. (doi:10.1080/00359190409519170) 79. AnAge database www.genomics.senescence.info

125

80. Morkel P vdB et al. 2012 Serial temperature monitoring and comparison of rectal and muscle temperatures in immobilized free-ranging black rhinoceros (Diceros bicornis). Journal of Zoo and Wildlife Medicine 43, 120–124. (doi:10.1638/2011-0009.1) 81. Lensch J. 1999 The two-humped camel (Camelus bactrianus). FAO Corporate Document Repository. ISSN 1014-6954. 82. Elgar MA, Harvey PH. 1987 Basal metabolic rates in mammals: allometry, phylogeny and ecology. Functional Ecology 1, 25. (doi:10.2307/2389354) 83. Holcomb KE. 2012 Behavioral and physiological responses of domestic horses to provision of shade in a hot, sunny environment. PhD Thesis. University of California, Davis. 84. Grubb P. 1981 Equus burchelli. Mammalian Species, 1. (doi:10.2307/3503962) 85. Fei Y, Hou R, Spotila JR, Paladino FV, Qi D, Zhang Z. 2016 Metabolic rates of giant pandas inform conservation strategies. Scientific Reports 6. (doi:10.1038/srep27248) 86. Jin Y, Qiao Y, Liu X, Pu T, Xu H, Lin D. 2016 Immobilization of wild giant panda (Ailuropoda melanoleuca) with dexmedetomidine–tiletamine–zolazepam. Veterinary Anaesthesia and Analgesia 43, 333–337. (doi:10.1111/vaa.12301) 87. Kinahan AA, Inge-moller R, Bateman PW, Kotze A, Scantlebury M. 2007 Body temperature daily rhythm adaptations in African savanna elephants (Loxodonta africana). Physiology & Behavior 92, 560–565. (doi:10.1016/j.physbeh.2007.05.001) 88. Adam D et al. 2016 Surgical implantation of temperature-sensitive transmitters and data- loggers to record body temperature in koalas (Phascolarctos cinereus). Australian Veterinary Journal 94, 42–47. (doi:10.1111/avj.12393) 89. Han H, Wei W, Nie Y, Zhou W, Hu Y, Wu Q, Wei F. 2016 Distinctive diet-tissue isotopic discrimination factors derived from the exclusive bamboo-eating giant panda. Integr. Zoo. 11, 447–456. (10.1111/1749-4877.12208) 90. Bocherens H, Koch PL, Mariotti A, Geraads D, Jaeger J-J. 1996 Isotopic Biogeochemistry (13C, 18O) of Mammalian Enamel from African Pleistocene Hominid Sites. Palaios 11, 306– 318. (doi:10.2307/3515241) 91. Fraser RA. 2005 A study of stable carbon, nitrogen and oxygen isotopes in modern Australian marsupial herbivores, and their relationships with environmental conditions. Australian National University, Canberra. 92. Harmon LJ, Weir JT, Brock CD, Glor RE, Challenger W. 2007 GEIGER: investigating evolutionary radiations. Bioinformatics 24, 129–131. (doi:10.1093/bioinformatics/btm538) 93. White CR, Seymour RG. 2003 Mammalian basal metabolic rate is proportional to body mass2/3. Proceedings of the National Academy of Sciences, USA 100, 4046-4049.

126

Appendix B

Supplementary information for Chapter 2

Sample selection Our study includes specimens from western Amazonia in Peru, considered Earth’s most biodiverse place for continental mammals (1). Using the list of mammals from Manu National Park (2), a well-known biodiversity hotspot, as an initial taxonomic guide, we sampled both Manu mammals and expanded the range of localities to include Tambopata National Reserve and a few other additional localities to augment the samples available from museum collections (increasing species diversity and numbers of individuals per species in our study). Virtually all samples come from national reserves, pristine and undisturbed areas in western Amazonia in Peru. To minimize heterogeneity of the ecosystems and restrict sampling to only lowland rainforest, we narrowed our search to sampling only specimens from sites <700 m elevation, so as to represent only lowland closed-canopy rainforests (with 4 additional specimens from a higher elevation cloud rainforest to augment taxon sampling/specimen numbers: Potos flavus [MUSM 19588], Dinomys branickii [MUSM 19855 and MUSM 7887], and Lagothrix lagotricha [MUSM 15905]). None of these additional samples differ significantly in d13C values from those of lower elevations. To maximize the phylogenetic diversity and range of body masses sampled, we included numerous families/orders across almost three orders of magnitude in body size (from 0.3 kg to 200 kg, including the largest extant terrestrial mammal species in Amazonia, Tapirus terrestris). Although there are a number of rodent and didelphimorph marsupial species below 0.3 kg in western Amazonia, sampling these taxa was unfeasible given the extremely small size of their teeth (which would have required the destruction of several entire teeth, which is not permitted by the museums housing these rare specimens). Our dataset therefore is more limited in sampling mammals <300 gr. Aiming to include all available relevant data from the literature from other western Amazonian sites, two specimens of Tapirus terrestris, from Colombia and Bolivia also have been included (3). Most samples were taken from late erupting/growth teeth (M3/m3, M2/m2, P4/p4, canines in carnivores), or ever-growing teeth in rodents (e.g., incisors), and xenarthrans (molariforms).

Stable isotope analyses Powdered samples of dental bioapatite were analyzed for d13C. Samples were first treated for 30 minutes with 3% H2O2 to remove organic contaminants, rinsed three times with distilled water, bathed for 30 minutes in 0.1 M Na buffered acetic acid to remove any secondary carbonates, rinsed three more times with distilled water, and placed in a low temperature oven to dry (protocol follows (4)). Samples were run on a Thermo Finnigan Gas Bench-CO. Hair samples were washed for 15 min in a 2:1 chloroform:methanol mixture to remove lipids and other surface contaminants, followed by two rinses with methanol, and a final 20 minute ultrasonic bath in DI water. Cleaned hair samples were dried for 24 hours in a low temperature convection oven (60 ºC) and then analyzed for d13C and d15N in an EA-IRMS (Delta V, Thermo Scientific) operated in a continuous flow mode. All the samples were processed and analyzed at the Stable Isotope Research Facility (http://sirfer.utah.edu), University of Utah, during a research in residence conducted by the corresponding author.

127

Analytical precision and accuracy of standards: Bioapatite (Internal reference materials: Carrara, LSVEC, and Marble) Precision: d13C = 0.2‰. Accuracy: d13C = -0.01‰ For each run (67 samples), 28 standards were included.

Hair (Internal reference materials: 2 glutamic acids, and ground bovine muscle): Precision: d13C = 0.1‰; d15N = 0.1‰. Accuracy d13C: -0.07‰; d15N = 0.04‰. For each run (24 samples), 9 standards were included.

13 Values are reported using the conventional permil (‰) notation wherein d C = (Rsample/Rstandard – 1) x 1000), and R=13C/12C. Values of d13C are reported relative to the VPDB standard. d15N values are reported relative to AIR. Isotope enrichment values (e*) of herbivores were calculated using body-mass dependent equations (5), as described in the main text. The superscript (*) 13 indicates that isotopic equilibrium is not assumed. For herbivores, d Cdiet refers to the predicted vegetation on which they feed. For secondary consumers (i.e., carnivores, insectivores, and 13 13 piscivores), d Cdiet refers to the predicted d C of the prey. We applied e*predator-prey values of - 1.3‰ for carnivores (6), and 7.8‰ for piscivores (average e*predator-prey value of 6 marine piscivorous mammals (7)). To our knowledge, not a single experimental e*predator-prey value exists for insectivores. With this limitation, we used the theoretical e*diet-bioapatite of 8‰ for the insectivore-insect e*predator-prey. This 8‰ value is the expected offset at 35°C for CaCO3, a value - mainly determined by the thermodynamics of equilibrium between blood HCO3 and blood dissolved CO2 (8, 9). However, acknowledging that future experimental studies might show that the reconstructed insectivore diets using this e*predator-prey may not be accurate, our relevant 13 conclusions have not included the insectivore d Cdiet data.

13 13 The d Cdiet was estimated from d C of dental enamel for all mammals with the exception of 13 xenarthrans. Among them, the d Cdiet of sloths (Bradypus variegatus and Choloepus hoffmanni) was estimated from outer dentine, whereas for edentulous anteaters (Cyclopes didactylus, Myrmecophaga tridactyla, Tamandua tetradactyla), and tiny-toothed armadillos (Dasypus 13 13 novemcinctus, Priodontes maximus) it was calculated from d Cbone. The d Cenamel data from 13 d Cbone were obtained using regression equations from a separate analysis of matched samples.

0 A 0 B

-5 -5

-10 -10 enamel treated

C enamel treated 13

d13C enamel treated enamel d13C d13C enamel treated enamel d13C d C 13 -15 -15 d -20 -20

-20 -15 -10 -5 -15 -10 -5

d13C bone treated 13d13C bone untreated d13C bone treated d C bone untreated 128

13 13 13 13 d Cenamel treated= 1.21 + 0.97*(d Cbone treated) d Cenamel treated= 0.44 + 1.04*(d Cbone untreated)

2 2 R : 0.74 R : 0.8 F-statistic: 205.7 on 1 and 73 DF, p-value: < 2.2e-16 F-statistic: 179.5 on 1 and 46 DF, p-value: < 2.2e-16 13 13 Figure S1. (A) Regression of treated d Cenamel values vs treated d Cbone values of matched 13 13 samples. (B) Regression of treated d Cenamel values vs untreated d Cbone values of matched samples. The correlation is strong in both, but even higher in (B).

Of the 168 enamel samples included, 24 were untreated (samples differentiated with “T” and “U” in the supplementary Excel file). We confidently included these untreated enamel samples because our experimental analysis of matched treated vs untreated enamel provided a virtually perfect correlation between individual test samples (Fig. S2). Similar results (supported by even larger datasets) also were obtained by (4) and (10).

0 13 13 d Cenamel untreated= 0.04 + 1.03*(d Cbone treated)

R2: 0.996

-5 F-statistic: 2800 on 1 and 11 DF, p-value: 1.352e-14

-10

enamel treated

d13C enamel treated enamel d13C

C 13 -15 δ

-20

-20 -15 -10 -5 0

δ13Cd13C enamel enamel untreated untreated

13 13 Figure S2. Regression of treated d Cenamel values vs untreated d Cenamel values of matched samples.

129

13 Figure S3. d Cdiet (calculated from dental bioapatite) for all western Amazonian mammals (A), and equatorial African mammals (B) analyzed. Boxplots represent the distribution of the data as 13 displayed on the upper left corner. Red dots represent the data per taxon. The d Cdiet of the herbivores represent the vegetation on which they feed. Taxa inside blue boxes are secondary

130

13 13 consumers, so their d Cdiet value represents the predicted d C1750 of the prey on which they feed. Numbers below (or above) each boxplot represent the number of samples analyzed per taxon.

Statistical Analyses presented in the main text

Statistical analyses, within and between orders, across continents, and between dietary guilds, were conducted with both parametric (t-test, ANOVA) and non-parametric (Mann-Whitney, Kruskal-Wallis) tests for significance. Normality was evaluated with the Shapiro-Wilk test. Unless noted, we report differences as statistically significant when p-values of pairwise comparisons for both parametric and non-parametric tests are £0.01.

13 The d Cdiet of herbivorous mammalian communities in western Amazonia (p-value = 3.512e-16) 13 and equatorial Africa have a non-normal distribution (p-value = 8.531e-10). The mean d Cdiet of western Amazonia and equatorial African herbivores are not significantly different at the 0.01 level, but differ significantly at the 0.05 level (p-value = 0.03).

Figure S4. Density and Q-Q plots for Amazonian (A) and African (B) mammalian herbivores.

131

13 The d Cdiet of carnivorous terrestrial carnivorans (i.e., Leopardus, Atelocynus, Galictis, Puma; 13 13 median d Cdiet =-13.48‰) vs the frugivorous carnivoran (Potos; median δ Cdiet =-27.42‰) is significantly different (p-value=0.001).

13 The d Cdiet of artiodactyls from Amazonia (mean= -25.04‰, median= -24.86‰) and equatorial Africa (mean= -25.57‰, median= -25.75‰) are not significantly different (p-value= 0.3). The 13 d Cdiet of primates from Amazonia (mean= -28.1, median= -28.2) and equatorial Africa (mean= -27.0, median= -27.0) are significantly different (p-value< 0.01), but the difference is only ~1‰.

15 13 d Nhair and hair-bioapatite d C enrichment 15 The d Nhair of primary and secondary consumers are significantly different (p-value= 0.0002)

Mean Plot with 95% CI p-value frugivore-carnivore 0.99

10 herbivore-carnivore 0.05 insectivore-carnivore 0.93

9 omnivore-carnivore 0.99 piscivore-carnivore 0.99 keratin - herbivore-frugivore <0.001 8 insectivore-frugivore 0.61 E*teeth-hair bioapatite *

e omnivore-frugivore 0.99 7 piscivore-frugivore 1 insectivore-herbivore 0.13

6 omnivore-herbivore 0.02 piscivore-herbivore 0.33 n=3 n=13 n=9 n=6 n=3 n=1 omnivore-insectivore 0.82 carnivore frugivore herbivore insectivore omnivore piscivore piscivore-insectivore 0.97 dietary categories piscivore-omnivore 1

Figure S5. Mean e*bioapatite-keratin values for each dietary category (black circles), sample numbers, 95% CI (blue bars), and Tukey pairwise comparison between group means (right box). Groups with significantly different e*bioapatite-keratin values (0.05 level) are in red.

132

Mean Plot Mean Plot A with 95% CI B with 95% CI 8.5 8.5 8.0

8.0 7.5 keratin - keratin - 7.0 7.5 bioapatite E*teeth-hair E*teeth-hair * bioapatite e * 6.5 e 6.0 7.0 5.5

n=3 n=22 n=10 n=22 n=13

omnivore primary secondary primary secondary dietary categories dietary categories

p-value primary-omnivore 0.47

secondary-omnivore 0.78

secondary-primary 0.73

Figure S6. (A) Mean e*bioapatite-keratin values for each dietary category (omnivore, primary consumer, secondary consumer; black circles), with sample numbers, 95% CI (blue bars) and Tukey pairwise comparisons between group means. No significant differences between dietary groups were found (lower box). (B) Mean e*bioapatite-keratin for each dietary category (primary and secondary consumers only, with omnivores binned in secondary consumers; black circles), with 95% CI (blue bars). No significant differences between groups were found in the primary- secondary (binned) consumer Tukey pairwise comparison (p-value= 0.27).

133

Mean Plot Mean Plot A with 95% CI B with 95% CI 6 4 4 2

keratin - keratin - diet 0 E*diet-hair E*diet-hair 0 * diet e * e -2 -2 -4

n=3 n=22 n=10 n=22 n=13

omnivore primary secondary primary secondary dietary categories dietary categories

p-value primary-omnivore 0.94 secondary-omnivore 0.002 secondary-primary <0.001

Figure S7. (A) Mean e*diet-keratin values for each dietary category (omnivore, primary consumer, secondary consumer; black circles), with sample numbers, 95% CI (blue bars) and Tukey pairwise comparisons between group means (lower box). No significant differences were found between omnivores and primary consumers, but secondary consumers were significantly different than primary consumers and omnivores (lower box). (B) Mean e*diet-keratin for each dietary category (primary and secondary consumers only, with omnivores binned in secondary consumers; black circles), with 95% CI (blue bars). Significant differences between primary and secondary (binned) consumers were found in Tukey pairwise comparison (p-value< 0.01).

134

13 Distribution and summary statistics of the d Cdiet values in other mammalian communities

13 a) d Cdiet of Pleistocene mammalian herbivores from Argentina (Fig. S9)

Min. 1st quartile Median Mean 3rd quartile Max. -25.4density plot of-23.2 late Pleistocene herbivores-21.6 Argentina -21.5 -20.2 -16.2

0.15 -15

0.10 -20 density Sample

0.05

-25

0.00

-24 -22 -20 -18 -16 d13Cdiet -30 13 δ Cdiet -2 -1 0 1 2 Theoretical

13 Figure S8. Median/mean and range (top), density plot (left) and Q-Q plot (right) for the d Cdiet of late Pleistocene mammals (n=103) from Argentina (11). Shapiro-Wilk normality test: p- value= 0.09.

13 b) d Cdiet of middle Miocene mammalian herbivores from Panama (Fig. S10)

density plot of middle Miocene herbivores Panama

0.15

-20

0.10

-25 density Sample

0.05

-30

0.00

-28 -26 -24 d13Cdiet -2 -1 0 1 2 Theoretical

13 d Cdiet 135

Min. 1st quartile Median Mean 3rd quartile Max. -29.3 -27.2 -26.0 -25.7 -23.8 -22.5

13 Figure S9. Median/mean and range (top), density plot (left) and Q-Q plot (right) for the d Cdiet of middle Miocene mammals (n=23) from Panama (12). Shapiro-Wilk normality test: p-value= 0.33.

Distribution and summary statistics of leaf d13C in closed canopy tropical rainforests

a) Leaf d13C from Rondonia, Brazil (Fig. S11)

Min. 1st quartile Median Mean 3rd quartile Max. -33.5 -31.3 -30.4 -30.5 -29.6 -27.2

d13C values of leaves, Rondonia, Brazil

0.3

-27.5

0.2

-30.0 Sample density

0.1 -32.5

-35.0 0.0 -2 -1 0 1 2 -32 -30 -28 Theoretical 13 δ Cd13C19701970 13 Figure S10. Median/mean and range (top), density plot (left) and Q-Q plot (right) for d Cleaves of closed canopy rainforest (n=139) in Rondonia, Brazil (13). Shapiro-Wilk normality test: p- value= 0.36.

b) Leaf d13C from Ituri, Congo (Fig. S12)

Min. 1st quartile Median Mean 3rd quartile Max. -34.9 -32.0 -29.4 -29.6 -27.8 -23.4

136

d13C values of leaves Ituri -20

-25 0.10

-30 Sample

density

0.05 -35

-40 0.00 -2 -1 0 1 2 Theoretical -35.0 -32.5 -30.0 -27.5 -25.0 d13C1970 13 δ C1970 13 Figure S11. Median/mean and range (top), density plot (left) and Q-Q plot (right) for d Cleaves of closed canopy rainforest (n=42) in Ituri, Democratic Republic of Congo (14). Shapiro-Wilk normality test: p-value= 0.55.

c) Leaf d13C from Venezuela (Fig. S13)

Min. 1st quartile Median Mean 3rd quartile Max.

-35.9d13C values-33.8 of leaves Venezuela -32.5 -31.8 -29.5 -25.4

-25

0.10

-30 Sample density 13 δ C1970 13 δ Cdiet 0.05 -35

-40

0.00 -2 -1 0 1 2 -36 -33 -30 -27 Theoretical 13 d13C1970 δ C1970 13 Figure S12. Median/mean and range (top), density plot (left) and Q-Q plot (right) for δ Cleaves of a closed canopy rainforest (n=36) in Venezuela (15). Shapiro-Wilk normality test: p-value= 0.08.

d) Leaf d13C from French Guyana (wet season) (Fig. S14)

Min. 1st quartile Median Mean 3rd quartile Max. -33.6 -32.5 -31.7 -30.9 -30.3 -26.1

137

d13C values of leaves Guyana (wet season)

-27.5 0.20

0.15 -30.0 density Sample 0.10

-32.5

0.05

-35.0

0.00

-32 -30 -28 -26 d13C1970 -2 -1 0 1 2 13 Theoretical δ C1970

13 Figure S13. Median/mean and range (top), density plot (left) and Q-Q plot (right) for d Cleaves of a closed canopy rainforest (wet season, n=15) of French Guyana (16). Shapiro-Wilk normality test: p-value= 0.02.

e) Leaf d13C in French Guyana (dry season) (Fig. S15)

Min. 1st quartile Median Mean 3rd quartile Max. -32.9 -32.1 -31.6 -30.3 -28.5 -25.1 d13C values of leaves Guyana (dry season)

0.15 -25

0.10 -30 Sample density

0.05 -35

0.00 -2 -1 0 1 2 -33 -31 -29 -27 -25 Theoretical 13d13C1970 δ C1970 13 Figure S14. Median/mean and range (top), density plot (left) and Q-Q plot (right) for d Cleaves of a closed canopy rainforest (dry season, n=15) of French Guyana (16). Shapiro-Wilk normality test: p-value= 0.02.

138

f) Leaf d13C from a subtropical monsoon forest, China (Fig. S16)

Min. 1st quartile Median Mean 3rd quartile Max.

-33.6d13C values-29.5 of leaves China -28.1 -24.1 -12.4 -9

0.075

0.050 density Sample -25

0.025

-50

0.000

-30 -25 -20 -15 -10 -2 -1 0 1 2 d13C1970 Theoretical 13 δ C1970 13 Figure S15. Median/mean and range (top), density plot (left) and Q-Q plot (right) for d Cleaves of a closed canopy rainforest (n=128) of a subtropical monsoon forest in China (17). Shapiro- Wilk normality test: p-value< 0.001.

139

15 Figure S16. d Nhair vs e*diet-bioapatite of Amazonian mammals. We note that the extremely low e*diet-keratin of carnivores (i.e., Galictis, Leopardus, and Atelocynus) is likely an artifact of a dietary reconstruction using a e*diet-bioapatite of -1.3 (6), based only on a small sample of hypercarnivores from North America.

140

1. Voss RS, Emmons LH (1996) Mammalian diversity in Neotropical lowland rainforests : a preliminary assessment. Bull Am Museum Nat Hist 230:1–115. 2. Solari S, Pacheco V, Luna L, Velazco PM, Patterson BD (2006) Mammals of the Manu Biosphere Reserve Mammals of the Manu Biosphere Reserve *. 0754(110):13–22. 3. DeSantis LG (2011) Stable isotope ecology of extant tapirs from the Americas. Biotropica 43(6):746–754. 4. Passey BH, et al. (2002) Environmental change in the Great Plains: an isotopic record from fossil horses. J Geol 110(2):123–140. 5. Tejada-Lara J V., et al. (2018) Body mass predicts isotope enrichment in herbivorous mammals. Proc R Soc B Biol Sci 285(1881). doi:10.1098/rspb.2018.1020. 6. Clementz MT, Fox-Dobbs K, Wheatley P V, Koch PL, Doak DF (2009) Revisiting old bones: coupled carbon isotope analysis of bioapatite and collagen as an ecological and palaeoecological tool. Geol J 44:605–620. 7. Clementz MT, Koch PL, Beck CA (2007) Diet induced differences in carbon isotope fractionation between sirenians and terrestrial ungulates. Mar Biol 151(5):1773–1784. 8. Friedman I, O’Neil JR (1977) Compilation of stable isotope fractionation factors of geochemical interest. 9. Salomons W, Mook WG (1986) Isotope geochemistry of carbonates in the weathering zone. Handbook of Environmental Isotope Geochemistry, pp 239–269. 10. Levin NE, Simpson SW, Quade J, Cerling TE, Frost SR (2008) Herbivore enamel carbon isotopic composition and the environmental context of Ardipithecus at Gona, Ethiopia. The Geology of Early Humans in the Horn of Africa: Geological Society of America Special Paper 446, eds Quade J, Wynn JG (Geological Society of America), pp 215–234. 11. Domingo L, Tomassini RL, Montalvo CI, Sanz-Pérez D, Alberdi MT (2020) The Great American Biotic Interchange revisited: a new perspective from the stable isotope record of Argentine Pampas fossil mammals. Sci Rep 10(1608):1–10. 12. MacFadden BJ, Higgins P (2004) Ancient ecology of 15-million-year-old browsing mammals within C3 plant communities from Panama. Oecologia 140(1):169–182. 13. Martinelli LA, et al. (1998) Stable carbon isotope ratio of tree leaves, boles and fine litter in a tropical forest in Rondonia, Brazil. Oecologia (114):170–179. 14. Cerling TE, Hart JA, Hart TB (2004) Stable isotope ecology in the Ituri Forest. Oecologia 138(1):5–12. 15. Medina E, Minchin P (1980) Stratification of δ 13C values of leaves in Amazonian rainforests. Oecologia 45(3):377–378. 16. Buchmann N, Guehl JM, Barigah TS, Ehleringer JR (1997) Interseasonal comparison of CO2 concentrations, isotopic composition, and carbon dynamics in an Amazonian rainforest (French Guiana). Oecologia 110(1):120–131. 17. Ehleringer JR, Lin ZF, Field CB, Sun GC, Kuo CY (1987) Leaf carbon isotope ratios of plants from a subtropical monsoon forest. Oecologia 72(1):109–114.

141