EARLY HOMININ ENVIRONMENTS IN SOUTHERN AFRICA:

A MICROMAMMALIAN PERSPECTIVE

by

JENNIFER NICOLE LEICHLITER

B.A, Colorado College, 2008

M.A. University of Colorado Boulder, 2011

A thesis submitted to the

Faculty of the Graduate School of the

University of Colorado Boulder in partial fulfillment

of the requirement for the degree of

Doctor of Philosophy

Department of Anthropology

2018

ii SIGNATURE PAGE

This thesis entitled: Early Hominin Environments in Southern Africa: A Micromammalian Perspective Written by Jennifer Nicole Leichliter has been approved for the Department of Anthropology

______Dr. Matt Sponheimer, Committee Chair

______Dr. Nico Avenant

______Dr. Herbert Covert

______Dr. Jaelyn Eberle

______Dr. Joanna Lambert

Date ______

The final copy of this thesis has been examined by the signatories, and we find that both the content and the form meet acceptable presentation standards of scholarly work in the above mentioned discipline.

iii ABSTRACT

Leichliter, Jennifer Nicole (Ph.D., Anthropology)

Early Hominin Environments in Southern Africa: A Micromammalian Perspective

Thesis directed by Professor Matt Sponheimer

Environmental change, especially the expansion of open C4 savanna grasslands at the expense of more heavily wooded C3 habitats, is frequently cited as a driver of hominin evolution.

The disappearance of the genus Australopithecus, the rise of the cranio-dentally robust

Paranthropus, and the emergence of Homo have all been linked to environmental change during the period from 3 and 2 Ma. The large mammal communities which co-existed with hominins are often used to reconstruct paleoenvironments during this period, but small mammals have been relatively underutilized despite their potential to provide information about hominin ecosystems at a fine scale. A primary goal of this dissertation was to determine whether patterns of evolutionary change observed in the large mammal fossil record of southern Africa during the

Plio-Pleistocene are also evident in the small mammal record.

A combination of stable carbon isotope analyses and taxon-dependent analyses were deployed, and in some cases significantly developed, in re-assessing the micromammalian fossil record associated with early hominins. Large gaps remain in our knowledge of small mammal isotopic ecology, especially in African ecosystems, where C4 plants are abundant and many landscapes relevant to hominin evolution persist. This dissertation focuses, in part, on understanding the relationship between small mammal stable carbon isotope composition and their habitats in a modern southern African C4 savanna and applying this information to the fossil record. It was determined that small mammal d13C compositions record habitat composition. iv

However, insectivores appear to track proportions of C3/C4 vegetation better than rodents.

Preliminary analyses of the d13C composition of fossil tooth enamel of small mammals from

early hominin-bearing deposits in southern Africa suggest that C4 resources may have been more abundant in the past.

Taxonomic-dependent faunal analyses indicate no clear shift from more mesic, closed habitats to more open-grassy habitats between 3 and 1 Ma in southern Africa, but instead suggest that the environment in this region was predominantly open with patches of mesic C3 woodland habitat. Australopithecus africanus and Paranthropus robustus appear to have been associated with both habitat types, perhaps taking advantage of resource-rich C3 habitat patches whenever they were available.

v TABLE OF CONTENTS

1 Introduction ...... 1 Background ...... 1 Chapter Summaries ...... 9 2 Small mammal insectivore stable carbon isotope composition as a proxy for habitat ...... 12 Introduction ...... 12 Materials and Methods ...... 19 2.2.1 Field Collection Methods ...... 19 2.2.2 Stable Isotope Analyses ...... 22 Results ...... 24 Discussion ...... 30 Conclusion ...... 36 3 Stable carbon isotope ecology of small mammal tooth enamel from barn owl roosts in the Sterkfontein Valley ...... 37 Introduction ...... 37 Materials and Methods ...... 40 3.2.1 Isotopic Analyses ...... 46 Results and Discussion ...... 49 3.3.1 Incisor-Molar Pairs...... 49 3.3.2 Modern Roost Site Comparisons ...... 52 3.3.3 Relative Abundance for Modern Roost Sites ...... 58 3.3.4 Fossil Analyses ...... 62 Conclusions ...... 67 4 Micromammal Fossil Descriptions ...... 69 Introduction ...... 69 Site Background Information ...... 71 4.2.1 Kromdraai B ...... 71 4.2.2 Sterkfontein Member 4 ...... 73 4.2.3 Swartkrans Member 1 (Hanging Remnant) ...... 73 4.2.4 Gladysvale External Deposit ...... 74 Systematic Paleontology ...... 75 5 African Small Mammal Taxonomic Habitat Index ...... 101 Introduction ...... 101 Development of the taxonomic habitat index ...... 102 Evaluating the taxonomic habitat index ...... 107 5.3.1 Biozones ...... 107 5.3.2 Roost Sites ...... 112 6 Paleoecological Analyses ...... 120 Introduction ...... 120 6.1.1 Assumptions ...... 124 Materials and Methods ...... 126 6.2.1 Species Richness and Diversity ...... 127 6.2.2 Similarity Indices ...... 129 6.2.3 Taxonomic Ratios ...... 129 6.2.4 Correspondence Analysis ...... 130 vi 6.2.5 Taxonomic Habitat Indices ...... 131 Results and Discussion ...... 132 6.3.1 Rarefaction ...... 132 6.3.2 Shannon-Weiner Index, Evenness, Simpson’s Dominance ...... 134 6.3.3 Similarity Indices ...... 140 6.3.4 Taxonomic Ratios ...... 142 6.3.5 Correspondence Analysis ...... 144 6.3.6 Habitat Reconstructions using THI ...... 148 Discussion ...... 156 6.4.1 Swartkrans and Sterkfontein ...... 156 6.4.2 Environmental change through time ...... 157 6.4.3 Hominin habitat associations ...... 157 Conclusion ...... 160 7 Conclusion ...... 162 8 Bibliography ...... 168 A. Appendix ...... 194

vii LIST OF TABLES

Table 2.1: Characteristics of the three habitats sampled including vegetation type (in percentages), ground cover, and GPS coordinates for each site...... 21 Table 2.2: Ecological information for each of the three species (C. cyanea, C. mariquensis, and M. varius) included in this study (data from Skinner and Chimimba, 2005; Happold, 2013)...... 23 Table 2.3: Stable carbon isotope compositions for each site sampled. δ13C values reflect the entire small mammal insectivore community sampled at each site...... 25 Table 2.4: Stable carbon isotope compositions of each species and for each species by site...... 28 Table 3.1: Vegetation data for modern roost sites ...... 43 Table 3.2: Ecological data for small mammal taxa (Happold, 2013 and Skinner and Chimimba, 2005; Taxonomy after Skinner and Chimimba)...... 45 Table 3.3: Carbon and oxygen isotope data for paired molars and incisors ...... 50 Table 3.4: Summary statistics for carbon and oxygen isotope data by taxon for each modern roost and fossil site (Mean±SD (N))...... 55 Table 5.1: Niche model differences between Reed (2003) and this study ...... 107 Table 6.1: Information about the fossil sites included in this study ...... 122 Table 6.2: Standardization of individual rarefaction curves ...... 134 Table 6.3: Diversity Indices Summary ...... 138 Table 6.4: Results of t-test for Shannon Wiener and Simpson Indices ...... 139 Table 6.5: Taxonomic ratios excluding and including the genus Otomys ...... 144 Table A.1: Niche models for African small mammals ...... 194 Table A.2: Taxonomic information and genus abbreviations...... 198 Table A.3: Minimum Number of Individuals (MNI) for each fossil site by taxon...... 200 Table A.4: Minimum Number of Individuals (MNI) for each roost site by taxon...... 202 Table A.5: Relative abundance (%) for each fossil site by taxon...... 203 Table A.6: Relative Abundance (%) for each roost by taxon...... 205 Table A.7: Presence/Absence data for each biozone by taxon...... 206 Table A.8: Presence/Absence data for each fossil site by taxon...... 209 Table A.9: Presence/Absence data for each roost site by taxon...... 211

viii LIST OF FIGURES

Figure 2.1: Location of modern and fossil sampling sites in the Cradle of Humankind World Heritage Site, South Africa. Map Data: Google, DigitalGlobe...... 20 13 Figure 2.2: Box plot of insectivore d Chair values by site...... 25 Figure 2.3: Box plot of body mass by taxon...... 26 13 Figure 2.4: Regressions of d Chair against body mass by site and taxon. There is no correlation 13 between d Chair and body mass...... 27 Figure 2.5: Least Squares Means Plot showing the lack of interaction between site and species. 29 13 Figure 2.6: d Chair values by species and site showing the directional shift in C. mariquensis and 13 M. varius d Chair values from site to site...... 31 13 Figure 2.7: δ C values and percentage C3 vegetation (where 100% C3 = 0% C4). The dashed line that crosses the graph at a diagonal represents an idealized micromammalian “ecological integrator” (Ideal Integrator). A linear mixing model using known regional end member values for C4 and C3 vegetation plus a diet to hair fractionation of +1.0‰ was used to generate a regression for an idealized integrator. Regressions were generated for each C3 vegetation proxy (T=Trees, CC=Canopy Cover, T+F=Trees + Forbs, and CC+F=Canopy Cover + Forbs) using recorded data and insectivore mean δ13C values for each habitat. δ13C means for each site are indicated on the graph...... 33 13 Figure 2.8: δ Cenamel (VDPB) values from fossil and modern small mammal insectivores. For comparison, values have been adjusted from original laser method values to conventional acid method based upon Passey and Cerling (2006). Fossil values were adjusted by +0.5‰ (ε*laser- conv). Modern values were adjusted by +0.3‰ (ε*laser-conv) and +1.5‰ (fossil fuel effect). 13 Fossil δ Cenamel values are notably higher than modern enamel values...... 35 Figure 3.1: Locations of Modern Roost sites (trees) and Fossil sites (filled circles) included in this study. Map Data: Google, DigitalGlobe ...... 41 Figure 3.2: d13C for incisor-molar matched pairs by site and taxon...... 51 Figure 3.3: Box plots d13C by roost and taxon with images of the vegetation at each modern roost site...... 54 Figure 3.4: Box-and-whisker plots of δ13C values for each modern roost site including A) all measured taxa; B) members of the order Rodentia only; C) three dominant rodent taxa (Mastomys, Micaelamys, Otomys); D) two dominant generalist taxa (Mastomys, Micaelamys). In panels C and D, sites marked with an asterisk are significantly different from the others (p < 0.05)...... 56 Figure 3.5: Theoretical and observed means (both unadjusted and adjusted for relative abundance) for each modern roost site. In each panel from left to right Kimberley (KRS) (at 10 %), Malapa (MRS) (at 30 %), Gladysvale (GRS) (at 35 %). Percentage C3 axis has been truncated from 0-50 % for clarity. A) All taxa; B) Rodents only; C) Dominant Rodent Taxa (Micaelamys, Mastomys, and Otomys); D) Dominant Generalist Rodent Taxa (Micaelamys and Mastomys only)...... 61 ix Figure 3.6: Box-and-whisker plots of d13C by fossil site...... 63 Figure 3.7: Box-and-whisker plots of small mammal carbon isotope compositions by taxon for Gladysvale fossil (GV) and Gladysvale modern roost (GRS) sites...... 66 Figure 4.1: Upper dental nomenclature from Miller (1912), lower dental nomenclature from Misnne (1969). Abbreviations: acc, anterocentral cusp; abc, anterobuccal cusplet; pbc, posterior buccal cusplet; a-bucc, anterobucal cusp; pc, posterior cingulum...... 71 Figure 5.1: Distribution of major biome types (A) and biozones (B) in Africa...... 108 Figure 5.2: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for all Biozones based on the THI for all taxa, using presence/absence data. Biozones are organized in ascending order from left to right by percentage non-woody vegetation...... 109 Figure 5.3: Habitat spectra for proportions of five habitat categories for all Biozones based on the THI for all taxa, using presence/absence data. Biozones are organized in ascending order from left to right by percentage semi-arid habitat predicted...... 110 Figure 5.4: Habitat spectra for proportions of five habitat categories for all Biozones based on the THI for rodents only, using presence/absence data. Biozones are organized in ascending order from left to right by percentage semi-arid habitat predicted...... 111 Figure 5.5: Map of southern Africa showing locations of barn owl roosts included in this study. Map Data: Google, DigitalGlobe...... 112 Figure 5.6: Map of major biomes of South Africa. Map from Mucina and Rutherford (2011). . 113 Figure 5.7: Satellite images of; (A) Roost 3, (B) Roost 5, (C) Roost 27, and (D) Roost 19. Google, DigitalGlobe...... 114 Figure 5.8: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for all Roost sites based on the THI for all taxa, using presence/absence data. Roosts are organized in ascending order from left to right by percentage non-woody vegetation...... 115 Figure 5.9: Habitat spectra for proportions of five habitat categories for all roosts based on the THI for all taxa, using presence/absence data. Roosts are organized in ascending order from left to right by percentage semi-arid habitat predicted...... 116 Figure 5.10: Habitat spectra for proportions of five habitat categories for all roosts based on the THI for all taxa, weighted for relative abundance. Roosts are organized in ascending order from left to right by percentage semi-arid habitat predicted...... 117 Figure 5.11: Satellite images of Roost 6...... 118 Figure 6.1: Locations of the early hominin sites included in this study. Maps from Gibbons et al., (2002) and Stratford et al., (2014)...... 123 Figure 6.2: Individual rarefaction curves for all southern African fossil sites with estimated species richness and 95% confidence interval...... 133 Figure 6.3: Individual rarefaction curves for ST-M4 and KA with estimated species richness and 95% confidence interval...... 134 x Figure 6.4: Shannon Weiner Diversity Index (H’) and Simpson’s Index of Dominance (D) for Makapansgat and all Sterkfontein Valley fossil sites. Fossil sites are arranged in chronological order from oldest (left) to youngest (right)...... 136 Figure 6.5: Dendrogram based on Jaccard similarity index evaluating species composition based on presence/absence data. The results were bootstrapped (n=9999). Sites associated with Australopithecus africanus are indicated in blue. Sites associated with Paranthropus robustus are indicated in red...... 141 Figure 6.6: Dendrogram based on Bray-Curtis similarity index evaluating species composition based on relative abundance data. The results were bootstrapped (n=9999). Sites associated with Australopithecus africanus are indicated in blue. Sites associated with Paranthropus robustus are indicated in red...... 142 Figure 6.7: Taxonomic ratios including Gerbillinae:Murinae, Dendromurinae:Murinae, Soricidae:Murinae. A value of 1 indicates equal proportions of each taxon. Fossil sites are arranged in chronological order from oldest (left) to youngest (right)...... 143 Figure 6.8: Detrended correspondence analyses based on presence/absence data and including all biozones, roost sites, and fossil sites. Eastern African sites and biozones distributed on the right side of the plot and southern African biozones, roosts, and fossil sites are distributed on the left. More arid biozones and roosts plot in the upper half of the DCA while more mesic habitats plot in the lower half...... 145 Figure 6.9: Detrended correspondence analyses based on presence/absence data and including all southern African roost and fossil sites. Modern roosts are distributed on the right side of the plot and fossil sites are distributed on the left. Roosts in more arid habitats plot in the upper half of the DCA while more mesic habitats plot in the lower half...... 146 Figure 6.10: Correspondence analysis including southern African fossil sites only based on all taxa and weighted for relative abundance. WW deposits, GON, DR, MAK, and GVED plot separately from the rest of the fossil sites...... 148 Figure 6.11: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for all biozones and all fossil sites based on the THI for all taxa, using presence/absence data. Sites are organized in ascending order from left to right by percentage non-woody vegetation...... 149 Figure 6.12: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for previously studied fossil sites based on the THI for all taxa,weighted for relative abundance. Sites are organized chronologically from left to right...... 150 Figure 6.13: Habitat spectra with all five habitat categories for previously studied fossil sites based on the THI for all taxa, weighted for relative abundance. Sites are organized chronologically from left to right...... 151 Figure 6.14: Habitat spectra with dichotomous representation of woody versus non-woody vegetation including DR, GON, and GVED. Spectra are based on the THI for all taxa, weighted for relative abundance. Sites are organized chronologically from left to right...... 153 Figure 6.15: Habitat spectra with all five habitat categories including DR, GON, and GVED. Spectra are based on the THI for all taxa, weighted for relative abundance. Sites are organized chronologically from left to right...... 154 xi Figure 6.16: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for eastern and southern African fossil sites. Spectra are based on the THI for rodents only, weighted for relative abundance. Eastern African sites are on the left side of the figure and southern African sites are on the right. Sites are organized chronologically from left to right. . 155 Figure 6.17: Habitat spectra with all five habitat categories for eastern and southern African fossil sites. Spectra are based on the THI for rodents only, weighted for relative abundance. Eastern African sites are on the left side of the figure and southern African sites are on the right. Sites are organized chronologically from left to right...... 156 Figure A.1: Individual rarefaction curves for southern and eastern fossil sites included in this study with estimated species richness and 95% confidence interval...... 212 Figure A.2: Individual rarefaction curves for eastern fossil sites only with estimated species richness and 95% confidence interval...... 212 1 1 INTRODUCTION

BACKGROUND

Environmental change, especially the expansion of open C4 savanna grasslands at the expense of more heavily wooded C3 habitats, is frequently cited as a driver of diversification and extinction within the hominin lineage (Vrba, 1985, 1995; Cerling 1992; Coppens 1994; deMenocal 1995, 2004; Reed 1997; Trauth et al. 2005; Maslin and Christensen 2007). The disappearance of the genus Australopithecus, the rise of the cranio-dentally robust genus

Paranthropus, and the emergence of the genus Homo throughout Africa have all been linked to environmental change during the period from 3 and 1 Ma (Vrba, 1975, 1985; Coppens, 1994; deMenocal, 1995; Behrensmeyer et al., 1997; Reed, 1997; Potts, 1998a, b, 2013; Trauth et al.,

2005; Maslin and Christensen, 2007; Maslin et al., 2014).

The role of expanding savanna habitats in hominin evolution has long been suspected

(Dart, 1925) and is supported by many faunal studies (Vrba, 1975, 1985, 1995; Reed, 1997). The bulk of these analyses have utilized large mammalian fauna (e.g. bovids, suids, carnivores, primates, etc.) to reconstruct past climatic conditions and environments during this period (e.g.,

Vrba, 1975, 1985; Coppens, 1994; Reed, 1997). In particular, the work of Vrba (1974, 1975,

1976, 1985, 1988, 1995) has significantly influenced current interpretations regarding the role of environmental change in southern Africa during the Plio-Pleistocene. Vrba (1975, 1985) used data from modern African bovid communities to determine if any taxa were significantly associated with particular habitat types (Vrba, 1985). She found that members of the tribes

Alcelaphini (e.g. hartebeest, wildebeest, blesbuck) and Antilopini (e.g. gazelles and springbuck) regularly constitute more than half of the taxa present (>65%) in open grasslands, while in closed environs, these two tribes combined made up less than a third (<30%) of the bovid community 2 (Vrba 1974). Vrba then applied this metric, referred to as the Alcelaphini-Antilopini criterion

(AAC), to the bovid fossil record at early hominin sites in the Sterkfontein Valley in southern

Africa.

The Sterkfontein Valley is located in the north central part of modern day South Africa and has produced thousands of Australopithecus africanus, Paranthropus robustus, and some early Homo specimens in association with rich faunal assemblages. Australopithecus africanus has also been found at two sites outside of the Sterkfontein area; Makapansgat, located to the north, and Taung, located to the southwest (Dart, 1925, 1948). Vrba (1974, 1975, 1985) initially studied the bovid fossil record at three of these sites, Sterkfontein, Swartkrans, and Kromdraai.

She found low proportions of grazing Alcelaphini and Antilopini in the Australopithecus africanus-bearing deposits at Sterkfontein Member 4 (~ 2.6-2.0 Ma) (i.e. low AAC) but relatively higher proportions of browsing and mixed-feeding tribes. Similar studies of the fauna at Makapansgat, which is the oldest hominin-bearing site in southern Africa (~3 Ma), also indicate a more mesic, moderately wooded environment for Australopithecus africanus (Vrba,

1985; Rayner et al., 1993; Reed, 1997; Sponheimer et al., 1999). Conversely, Vrba found that grazing bovids and open habitats dominated the younger Sterkfontein Member 5 (~1.7 Ma),

Swartkrans Members 1-3 (1.8-1.0 Ma), and Kromdraai deposits (1.8 Ma) from which

Paranthropus robustus specimens and early Homo derive (Vrba, 1975, 1985).

Vrba’s analyses recorded a clear pattern—the bovid assemblages from older deposits

(associated with Australopithecus africanus) indicated more closed, mesic habitats than the bovid assemblages from younger deposits containing Paranthropus robustus and early Homo. In expanding her analyses to eastern African hominin fossil localities, Vrba found a similar pattern.

Moreover, she further identified distinct periods of intensified speciation in grazing and open- 3 habitat adapted taxa and extinction in browsing/mixed-feeding and closed-habitat adapted bovid lineages between 3 and 1 Ma. These “pulses” of speciation/extinction appeared to be correlated with similar events in several other groups, including hominins. From these data, Vrba inferred a directional shift in the prevailing environmental conditions from relatively closed to more open habitats through time. Vrba’s theory regarding the relationship between mammalian evolution and environmental change during the Plio-Pleistocene is referred to as the “Turnover Pulse

Hypotheses” and it has been enormously influential in the field of paleoanthropology (Vrba

1985, 1995; Behrensmeyer et al., 1997; deMenocal, 2004; Bobe et al., 2002; Bobe and

Behrensmeyer, 2004; Faith and Behrensmeyer, 2013; Potts, 2013).

Subsequent analyses have supported Vrba’s interpretations to greater or lesser degrees

(Reed, 1997; Potts, 1998; Maslin et al., 2014). Indeed, Africa appears to have become both cooler and drier over the course of the Plio-Pleistocene (deMenocal, 1995; Trauth et al., 2005;

Maslin and Christensen, 2007; Maslin et al., 2014). That these climate changes resulted in the expansion of open savanna grasslands and contraction and fragmentation of woodland habitats is, on a broad scale, well supported by multiple paleoenvironmental proxies (e.g. pollen analyses, soil carbonates, marine sediments, lake sediments, stable carbon isotope compositions of fossil teeth). For instance, stable isotopic analyses of soil carbonates (Levin et al., 2004; Wynn, 2004;

Ségalen et al., 2007; Levin, 2013) and fossil tooth enamel (Harris et al., 2008; Brachert et al.,

2010; Lee-Thorp and Sponheimer, 2007; Wynn et al., 2016) in eastern and southern Africa corroborate interpretations of paleoenvironments based on taxonomic studies, and suggest a shift towards open, grassland environments and an increasing abundance of C4 plants during the Plio-

Pleistocene (Lee-Thorp et al., 2003, 2010; Cerling, 2013; Wynn et al., 2016). Moreover, stable isotope analyses of early hominin taxa hint that the increasing use of C4 resources by hominins 4 may have played a role in hominin masticatory evolution and diversification (Sponheimer et al.,

2013).

Within the context of this narrative, it is inferred that the loss and fragmentation of mesic, wooded habitats resulted in the extinction of the genus Australopithecus and the evolution of hominin taxa adapted to open-habitats (i.e. Paranthropus and Homo). This is, of course, an overly simplistic representation of a far more complex and ongoing discussion (e.g., Bobe and

Behrensmeyer, 2004; Cerling et al., 2011). There is, for instance, significant debate regarding both the nature and timing of environmental change across regions, sub-regions, and localities

(Kingston, 2007; Potts 1998; Bobe and Behrensmeyer, 2004; Segalen et al., 2007; Maslin et al.,

2014). Embedded in this conversation regarding the broader environmental framework of hominin evolution is a more nuanced discussion about habitat use and preference in early hominin taxa (Sikes, 1994; de Ruiter et al., 2008). It has been suggested that some hominin taxa preferred certain microhabitats within savanna landscapes just as some modern primates do today. For example, ‘savanna’ chimpanzees (Pan troglodytes) at Fongoli in southeastern Senegal selectively use closed-canopy habitats such as gallery forests within a broader savanna mosaic environment (Pruetz et al., 2009; Sponheimer et al., 2013). With respect to early hominins,

Paranthropus has been argued to prefer edaphic grasslands with a wetland component (Reed,

1997), while Homo is often associated with more xeric environs (Shipman and Harris, 1988;

Stanley, 1992; McKee, 1991; Reed, 1997), although Paranthropus and early Homo apparently occupied the same broader environment. It has further been postulated that this segregation may have reduced niche competition between the two taxa (Stanley, 1992; Tattersal, 2009)

Not all lines of evidence support Vrba’s interpretation of southern African hominin habitat associations, however. Stable carbon isotope, microwear, and some faunal analyses, for 5 instance, suggest that it is overly simplistic. For example, de Ruiter et al. (2008) examined the fauna at several Paranthropus robustus fossil sites and found this species to be most strongly associated with woodland-adapted taxa. Moreover, they found a significant negative correlation between Paranthropus robustus and open-grassland adapted taxa – that is, as the abundance of individuals belonging to grassland-adapted mammalian taxa increases in Paranthropus-bearing deposits, the number of hominins decreases. The authors argue that Paranthropus robustus was likely a habitat generalist rather than an open grassland specialist. Interestingly, stable carbon isotope and microwear analyses of Paranthropus robustus teeth indicate that this taxon was also a dietary generalist with a seasonally varied diet consisting of C3 and C4 resources most likely obtained by foraging in both open and closed habitats (Sponheimer et al., 2006). These findings stand in contrast to δ13C data from eastern African Paranthropus species (i.e. P. boisei, P. aethiopicus) which appear to have predominantly consumed C4 resources (van der Merwe et al.,

2008; Cerling et al., 2011, 2013; Sponheimer et al., 2013). Furthermore, stable carbon isotope compositions for Australopithecus africanus and Paranthropus robustus indicate that both taxa consumed a mix of C3 and C4 resources, suggesting that neither species foraged exclusively in one habitat type or another (Sponheimer and Lee-Thorp, 1999; Sponheimer et al., 2006; van der

Merwe et al., 2008; Ungar and Sponheimer, 2011).

Most significantly for the current study, Avery (2001) found little evidence for environmental change between Sterkfontein Member 4 and Sterkfontein Member 5/Swartkrans

Members 1-3 in her analyses of the small mammal fossil assemblages at these sites. In her reconstruction of the paleoclimate for these deposits, Avery argues that hotter and drier conditions (<500mm rainfall per annum), and more seasonal rainfall, prevailed during the Plio-

Pleistocene than characterize the area today. She finds little evidence of any significant 6 differences in the microfaunal communities between the older and younger deposits. Clearly, this interpretation is at odds with the findings of Vrba (1975, 1985) and others, which identify a major shift from more mesic conditions for Sterkfontein Member 4 and Makapansgat Member 3, to drier, open grassland conditions (though not exceeding conditions known in the area today) at

Sterkfontein Member 5, Swartkrans Members 1-3, and Kromdraai (Vrba, 1985; Reed, 1997; Lee-

Thorp et al., 2007). Considering this discrepancy, Avery’s results warrant further investigation, yet little work has been done to verify her findings. In addition, micromammalian records from southern and eastern African hominin sites have rarely been compared to determine if similar patterns are evident in both (but see Denys, 1999).

Small mammals are well-suited to address questions about the nature, magnitude, and timing of environmental change during early hominin evolution for several reasons. For one, most modern African genera were fully established by ~ 5mya, inviting direct comparison between Plio-Pleistocene small mammal fauna and extant groups and strengthening interpretations based upon ecological similarities between closely allied fossil and modern taxa

(i.e. taxonomic uniformitarianism) (De Graaff, 1961; Denys, 1999; Skinner and Chimimba,

2005; Winkler et al., 2010; Happold, 2013).

Considered together, the ‘small mammals of Africa’ (including the orders Afrosoricida,

Chiroptera, Erinaceomorpha, Lagomorpha, Macroscelidea, Rodentia, and Soricomorpha), make up 74.1% of the mammalian taxonomic diversity on the continent – the order Rodentia alone accounts for 36% (Happold, 2013). They are ecologically diverse, occupying even the most extreme environments in Africa and many groups show habitat specificity and dietary selectivity

(Skinner and Chimimba, 2005; Happold, 2013). 7 Finally, most micromammals are highly spatio-temporally constrained. They have small home ranges (generally <1km) and short lifespans and are therefore very much tethered to their local environmental conditions (Skinner and Chimimba, 2005; Happold, 2013). Unlike large fauna, which can weather lean years and range broadly across the landscape in times of resource stress, small mammals must face conditions directly, either adapting or going locally extinct.

This dissertation focuses specifically on non-volant ‘micromammals’ which have a body mass of

≤ 300g.

Despite their potential, relatively few studies have applied micromammalian datasets to questions about early hominin paleoecology in the same manner that the macrofaunal record has been employed. This is, in part, because large mammals are more easily studied, and thus more is known about their ecology and fossil history. In addition, large mammal remains are relatively easier to collect and identify at fossil sites, whereas labor intensive methods such as sieving are often required for recovery of small mammal remains (this the case at many eastern African sites). Fortunately, this is not an issue in the southern African deposits where acid preparation of fossil breccias produce large amounts of small mammal material. Lastly, micromammals have also been underutilized because appropriate technology was not yet available for the application of methods like stable isotope and microwear analyses to very small fossil teeth.

The overarching goal of this dissertation is to begin filling gaps in micromammalian datasets so that they can be more readily compared to macrofaunal analyses. Specifically, a combination of stable carbon isotope analyses and taxon-dependent faunal analyses were deployed, and in some cases significantly developed, as a first step towards re-assessing the micromammalian fossil record associated with early hominins in southern Africa. 8 13 With δ C values for most species of hominin now known, it is clear that C4 resources played a significant role in early hominin diets and C4 environments likely shaped hominin masticatory anatomy (Sponheimer and Lee-Thorp, 1999; Sponheimer et al. 2006, 2013; Ungar and Sponheimer, 2011; Grine et al., 2012). For this reason, understanding how C4 resources were distributed on the landscape at various times and places, how environments changed over time, and how this influenced hominin macorevolution is of great interest to paleoanthropologists.

Stable isotope analyses of carbon in fossil tooth enamel from the fauna which co-existed with early hominins provide a useful line of supplementary evidence regarding aspects of early hominin environments and environmental change throughout the Plio-Pleistocene. Stable isotope analyses target questions about fossil taxa dietary ecology, which in turn encodes information about habitat use in ancient animals and the environmental conditions that prevailed at various points in the past (Luyt and Lee-Thorp 2003; Sponheimer and Lee-Thorp 2003; Lee-Thorp et al.

2007). Habitat reconstructions using information about fossil taxa dietary ecology are predicated upon the relatively simple principle that animals eat what is readily and sufficiently available in their environment (Lee-Thorp et al., 2007).

As is the case with taxonomic analyses, the stable isotope analyses of Plio-Pleistocene fauna in Africa have focused heavily on large mammals (e.g. bovids), while smaller fauna, particularly micromammals such as rodents, bats, and shrews, have not been well sampled. Thus, while we know a great deal about dietary/paleodietary variation and habitat preference in large fauna, we know relatively little about these variables from an isotopic perspective in small fauna.

We do not know, for example, whether modern African small mammal faunas vary isotopically according to their habitat. Without such basic information we cannot assume that the dietary 9 ecology of small fauna can be used as an environmental proxy in the same manner as it can in large fauna.

Taxon-dependent faunal analyses of micromammalian datasets in eastern Africa generally support the findings of macrofaunal analyses at hominin-bearing sites (i.e. Jaeger,

1976; Wesselman et al., 1985; Fernandez-Jalvo et al., 1998; Reed, 2003, 2007, 2011) and corroborate hypotheses that open habitats expanded between 3 and 1 Ma. However, Avery’s

(2001) assessment of the micromammal assemblages at Sterkfontein and Swartkrans conflict with macrofaunal data from the same deposits. It therefore remains unclear how southern African microfaunal and macrofaunal records speak to one another. Determining whether similar patterns of environmental change are evident in the large and small mammal record of southern Africa is a primary goal of the current study.

CHAPTER SUMMARIES

The data presented in Chapters 2 and 3 represent a first attempt to address this gap in our knowledge regarding the relationship between small mammal stable isotope compositions and the habitats in which they are found. The goal of both studies was to determine how small mammal stable carbon isotopes vary at intraspecific, interspecific, and community levels with respect to habitat. The studies were conducted using specimens collected in the Cradle of

Humankind, World Heritage Site, South Africa, which encompasses all of the Sterkfontein

Valley fossil deposits. In Chapter 2, stable carbon isotope analyses of hair from small mammals live-trapped in a variety of habitat types are presented. This chapter focuses specifically on the stable isotope compositions of small mammal insectivores, which are hypothesized to act as ecological ‘integrators,’ and are therefore potentially useful proxies for community δ13C composition. 10 In Chapter 3, stable carbon isotope analyses of tooth enamel in small mammal taxa from modern barn owl (Tyto alba) roosts in the Cradle of Humankind are presented. Material from barn owl roosts was utilized for several reasons; 1) large samples of small mammal remains can be easily collected at roosting sites, 2) barn owls have been shown to be efficient and relatively unbiased samplers of the small mammal communities in their hunting ranges (Avery, 1998;

Avenant, 2005; Aven ant and Cavallini, 2007; Terry, 2010), and 3) the micromammal fossil assemblages which are the focus of these analyses are argued to have been accumulated primarily by barn owls (Davis, 1959; De Graaff, 1960; Brain, 1981; Pocock, 1985, 1987;

Andrews, 1990; Avery, 1998, 2001, 2010). Thus, stable isotope analyses of material from barn owl roosts offers the additional benefit of taphonomic control, facilitating application to the fossil record. Preliminary results of stable isotope analyses of fossil small mammal tooth enamel samples from Sterkfontein Member 4 (~2.5 Ma), Swartkrans Member 1 – Hanging Remnant

(~1.8 Ma), and Gladysvale (~700 Ka) are presented at the ends of Chapters 1 and 2.

Faunal analyses are presented in Chapters 4-6. In Chapter 4, new micromammal material from Sterkfontein Member 4, Swartkrans Member 1 (Hanging Remnant), Kromdraai B, and

Gladysvale External Deposits is described. These identifications are incorporated into the faunal datasets used in the paleoecological analyses presented in Chapter 6. These new materials significantly supplement the data available for analysis at deposits with small sample sizes.

Chapter 5 focuses on the development of a small mammal taxonomic habitat niche model for application to the fossil record. A Taxonomic Habitat Index (THI) is a cumulative index obtained by combining the habitat indications of all species contained in a community or assemblage (Nesbit-Evans et al., 1981; Fernandez-Jalvo et al., 1998; Reed, 2003, 2007). In essence, THI is a method of aggregating the habitat associations of all micromammalian taxa 11 present in an assemblage to produce a composite interpretation of the paleoenvironment (Reed,

2003; Nel and Henshilwood, 2016; Nel et al., 2017). If designed appropriately, this cumulative index should indicate the dominant habitats characterizing each assemblage. THIs have been employed in studies of Plio-Pleistocene micromammal assemblages in eastern Africa but have not yet been applied to contemporaneous southern African assemblages. The THI employed here was specifically developed for this study both for application to the southern African micromammalian record and to facilitate direct comparison between the two primary regions where early hominin fossils have been found.

Finally, paleoecological reconstructions using the micromammal assemblages from several hominin-bearing deposits in southern Africa are presented in Chapter 6. The goals of this chapter include 1) incorporating additional micromammalian data from new and existing Plio-

Pleistocene sites into paleoecological analyses, 2) evaluating the taxonomic composition, diversity, and taxonomic habitat index at each site, 3) re-evaluating Avery’s (2001) findings at

Sterkfontein and Swartkrans, and 4) investigating whether differences in habitat association in

Australopithecus africanus versus Paranthropus robusts and early Homo are indicated in the micromammal record. 12 2 SMALL MAMMAL INSECTIVORE STABLE CARBON ISOTOPE COMPOSITION AS A

PROXY FOR HABITAT

INTRODUCTION

Understanding the environmental context of hominin evolution is a primary focus of paleoanthropological research (e.g. Dart, 1925; Vrba, 1975, 1985, 1988; Reed, 1997; Potts 1998;

2013; Bobe et al., 2002; Bonnefille et al., 2004; Kingston, 2007; Maslin et al., 2015). Though complicated and not unidirectional, there is strong evidence that during the Plio-Pleistocene, large-scale tectonic and climatic shifts influenced African ecosystems, driving the expansion of open environments at the expense of forested habitats (Vrba, 1975; deMenocal, 1995, 2004;

Reed, 1997; Potts, 1998; Trauth et al., 2003, 2005, 2007, 2009; Kingston, 2007; Maslin and

Christensen, 2007; Maslin and Trauth, 2009; Reed et al., 2014; Maslin et al., 2015). Against this larger backdrop of continental change, a wide diversity of mosaic habitats emerged that many believe played an important role in hominin evolution (Dominguez-Rodrigo, 2014; Reynolds et al., 2015).

Carbon isotopic analysis of paleosol carbonates is a well-established tool in paleoenvironmental reconstruction (Cerling, 1992, 1999; Cerling and Hay, 1986; Cerling et al.,

1989, 1991, 1997, 2011; Levin et al., 2008, 2011; Segalen et al., 2007; Sikes, 1994, 1995; Wynn et al., 2000, 2004). Such analyses are based upon the different photosynthetic pathways used by most trees, shrubs and forbs (C3 plants) and tropical grasses and sedges (C4 plants). Plants using

13 13 12 C3 photosynthesis discriminate against C resulting in a lower ratio of C to C in plant tissue

13 relative to that of atmospheric CO2. C4 plants discriminate less against C, and therefore exhibit a relatively higher ratio of 13C to 12C. This differential discrimination results in distinct, non- overlapping carbon isotope distributions across the two types of plants (Smith and Epstein, 13 1971). Carbon isotopes are of particular utility in the savannas of Africa, where the landscape consists of a mix of C3 and C4 vegetation, distributed across the two major functional vegetation types in savanna environments (C3 trees and shrubs; C4 grasses). The carbon isotopic composition of paleosol carbonates reflects that of vegetation growing in soil during the time of soil formation (Cerling, 1984, 1999; Cerling et al., 1989, 1991; Quade et al., 1989; Cerling and

Quade, 1993).

The carbon isotopic composition of foods is incorporated into the tissues (e.g. bone, hair, enamel) of the animals that eat them (DeNiro and Epstein, 1978; Vogel, 1978; Cerling and

Harris, 1999). Thus, the ratio of 13C / 12C in body tissues reveals information about the relative contribution of C3 and C4 foods to diet. For example, isotopic differences in tissue composition have been used to discriminate grazing (C4 consuming), browsing (C3 consuming), and mixed- feeding (mixed C3 / C4 consuming) herbivores in both modern and fossil assemblages (Ambrose and DeNiro, 1986, Koch et al., 1994, 1998; Cerling and Harris, 1999; Zazzo et al., 2000; Luyt and Lee-Thorp, 2003; Sponheimer and Lee-Thorp, 2003; Cerling et al., 2003; Kingston and

Harrison, 2007; Lee-Thorp et al., 2007; Levin et al., 2008; Bedaso et al., 2010, 2013). Diet, in turn, reveals information about habitat since an animal can only consume what is available in its environment. For example, if C4 grasses heavily dominate a landscape, and C3 resources are rare, it is likely that many animals foraging in that environment will consume C4 foods given that they are the most abundant and readily available resource. Thus, by using carbon isotope analysis to reconstruct the diets of communities of herbivores, one can make inferences about the dominant vegetation on a given landscape.

Carbon isotope analyses of tooth enamel have been widely employed by paleoanthropologists to understand past habitats (Luyt and Lee-Thorp, 2003; Sponheimer and 14 Lee-Thorp, 2003; Kingston and Harrison, 2007; Lee-Thorp et al., 2007; Levin et al., 2008;

Bedaso, 2010, 2013). Such studies use a variety of analytical approaches based on both intra- specific comparisons (e.g., within Bovidae) (Sponheimer, 1999; Luyt and Lee-Thorp, 2003;

Sponheimer and Lee-Thorp, 2003) and comparisons across multiple taxonomic groups, specific dietary guilds, and whole ecological communities (Zazzo et al., 2000; Kingston and Harrison,

2007; Levin et al., 2008; Sponheimer and Lee-Thorp, 2009; Bedaso et al., 2010, 2013). These

13 studies use d C values to determine percentages of dietary categories (e.g., C3 consumers, C4 consumers) within assemblages (Luyt and Lee-Thorp, 2003; Sponheimer and Lee-Thorp 2003;

Lee-Thorp et al., 2007), or use d13C distribution or summary data for a site to compare with other modern or fossil datasets (Kingston and Harrison, 2007; Levin et al., 2008; Sponheimer and Lee-

Thorp, 2009), while some incorporate relative abundance data or biomass calculations (Luyt and

Lee-Thorp, 2003; Sponheimer and Lee-Thorp, 2003, 2009; Lee-Thorp et al., 2007; Bedaso et al.,

2010, 2013).

In practice, it is usually not difficult to distinguish C3 forests from environments where

C4 grasses are present (Ambrose and DeNiro, 1986; Cerling et al., 2003; MacFadden and

Higgins, 2004). Such environments range from woodland with some grass to treeless grassland, but using carbon isotopes to reconstruct specific habitats within this broad range of environments is complex (Sponheimer and Lee-Thorp, 2009). Such reconstructions are limited by a poor understanding of how carbon isotopes are distributed in modern foodwebs, both at the level of the individual species and at the community level (Gannes et al., 1997; Codron et al., 2007b).

For instance, how do we quantify the difference between “wooded grassland” and “grassland”

(sensu Reed et al., 2013)? What differences should be manifest isotopically in environments with

10%, 25%, or 50% tree cover? These distinctions are important because all African ecosystems 15 associated with fossil hominins contain a mix of C3 and C4 vegetation, and are often generalized as mosaics, although the relative abundance of C3 and C4 plants as well as the nature and distribution of plant types at many sites are heavily debated (Reynolds et al., 2015). Consensus in the field is that mosaic habitats require better characterization and more detailed study in modern context to evaluate their role in hominin evolution (Reynolds et al., 2015).

So how do we improve our understanding of how carbon isotopes are distributed in modern African ecosystems, and how they reflect local vegetation? Since it is often impractical to sample an entire ecological community, much less incorporate data on the relative abundance of each taxon (especially given taphonomic considerations in fossil assemblages), an ideal approach would be to identify a reliable isotopic ecological “integrator” – an organism whose isotopic signature accurately and consistently reflects the actual proportion of C3 and C4 vegetation in its habitat. This could be achieved by looking for taxa that are highly opportunistic, eclectic feeders that shift their diets in response to their ecological setting. Isotopically, most individual species are relatively poor indicators of environmental characteristics. The vast majority of African bovids are C3 and C4 specialists and maintain their specialized diets in heterogeneous environments (Sponheimer et al., 2003a; Cerling et al., 2003). Even flexible, generalist herbivores do not seem to reliably record environmental characteristics isotopically, instead tending to browse or graze preferentially. For instance, both nyala (Tragelaphus angasii) and impala (Aepyceros melampus) are classified as “mixed-feeders”, yet the former prefers browse and the latter grass (Codron et al., 2007b). Similarly, the diets of mixed-feeding African elephants (Loxodonta africana) do not always accurately reflect regional differences in grass biomass, tree density, or canopy cover (Koch et al., 1995; Codron et al., 2006, 2011). Even most 16 rodents, which are widely considered generalist feeders, were not found to change their diets across different habitats within a savanna environment (Codron et al., 2015).

Traditionally, organisms at lower trophic levels, including primary consumers such as herbivores, have been used to make inferences about habitat under the assumption that their direct interaction with vegetation reflects the distribution of C3 and C4 resources on a landscape.

However, it has been suggested that organisms at higher trophic levels might actually better reflect overall environmental characteristics by acting as ecological integrators, concentrating signals from multiple variable sources and averaging inputs from lower trophic levels through space and time (de Ruiter et al., 2005). Many carnivores are opportunistic dietary generalists, and select their prey primarily on the basis of body size, abundance and availability (Vezina, 1985;

Murray et al., 1994; Radloff and du Toit, 2004; Hayward and Kerley, 2008), all variables that are intrinsically linked to habitat. Isotopic sampling of predators increases effective “sample size”, because predators integrate the tissues of multiple herbivore species which themselves consume many different primary producers (Bump et al., 2007).

The argument that predators can serve as isotopic integrators has been tentatively supported by data from fossil carnivores (Lee-Thorp et al., 2000, 2007; Bump et al., 2007;

Hopley and Maslin, 2010). The idea has not been well tested in the context of modern isotopic ecology, but some support for its validity does exist. Bump et al. (2007) determined that environmental isotope patterns were better represented by higher trophic levels in both experimental and observational contexts. For example, using carbon isotope data they determined that wolves (Canis lupus) record historical trends in atmospheric CO2 more accurately than tree rings. Similarly, data from Codron et al. (2007a) suggests that South African savanna-dwelling carnivores differ in d13C based on feeding behavior and associated habitat 17 preferences of different prey types. In particular, lions foraging in open grasslands were found to have higher d13C values than lions associated with more wooded habitats.

However, large to medium bodied carnivores are capable of ranging significant distances in search of prey and likely sample from a variety of habitat types (Gittleman and Harvey, 1982).

While demonstrably useful in recording large-scale spatial and temporal patterns of carbon isotopes in the palaeontological record (e.g. changes in atmospheric CO2 levels) (Kohn et al.,

2005; Bump et al., 2007; Hopley and Maslin, 2010), they may offer poorer resolution at smaller spatial scales. In a paleoecological context, there is the added complication that predators are somewhat rare in the fossil record relative to other organisms (Brain, 1981; Werdelin and Lewis,

2005; Patterson et al., 2014).

Smaller bodied, insectivorous mammals such as Soricids (shrews) and Macroscelids

(sengis or elephant shrews) are ideal potential candidates for carbon isotopic studies. Unlike larger predators, micromammals have restricted home ranges and are therefore closely tied to local habitat (Skinners and Chimimba, 2005). Shrews, in particular, are common elements of almost all African ecosystems and live in a diverse range of habitat types (Churchfield, 1990;

Dickman, 1988, 1995; Churchfield et al., 1997a, 2004; Dudu et al., 2005; Skinner and

Chimimba, 2005; Happold and Happold, 2013). They feed almost exclusively upon invertebrate prey, eat a wide variety of insect taxa, and are highly opportunistic, switching prey according to availability (Churchfield, 1982a, 1990, 1995; Churchfield et al., 1991, 1997b). Their high metabolic rates necessitate almost constant foraging and they are voracious feeders (Churchfield,

1982a, b, 1990, 2002; Churchfield et al., 1991, 1997b). Churchfield and Brown (1987) estimated that the daily consumption rate of invertebrates by shrews in grassland amounted to 6,800 prey per hectare. In a meta-study of the relationships between predator and prey, Vezina (1985) 18 determined that insectivores generally take a wider range of prey size and type than do larger carnivores. This agrees well with data on the feeding ecology of sympatric African shrews.

Churchfield et al. (2004) and Dudu et al. (2005) found no evidence of prey size selection or dietary specialization in multi-species communities of west African forest shrews. Therefore, micromammalian insectivores have the potential to act as ecological integrators capable of providing environmental resolution at much smaller spatial scales than larger carnivores.

Micromammalian insectivores are rarely the focus of isotopic studies (but see Baugh et al. 2004), however some data do exist for African taxa (Hopley et al., 2006; Dammhann et al.,

2013; Symes et al., 2013; Van den Heuvel and Midgley, 2014). Of these studies only Symes et al. (2013) includes data from modern insectivores collected in mixed C3/C4 habitats. That study examined resource partitioning in sympatric small mammals (including two shrew species) in an

African forest-grassland mosaic and determined that specimens trapped in Afromontane grassland exhibited higher d13C values than those captured in Afromontane forest. In all studies, sample sizes for insectivorous taxa were very small. Given the dearth of available data on the topic, the degree to which the d13C values of insectivorous micromammals reflect habitat, particularly in mosaic savanna environments, remains unclear.

In this study, we attempt to address this gap by assessing the degree to which micromammalian insectivores record habitat signals in a mosaic environment under spatially and seasonally constrained conditions. We sampled three sympatric shrew taxa in three microhabitat types within a southern African savanna (see Codron et al., 2015 for isotopic composition of rodents in the area). Sampling sites were all located within 2 km of one another, but range from relatively open (less than 5% tree cover) to moderately closed (~ 50% tree cover). We compared insectivore d13C values between microhabitat types, and across taxa, in order to test whether 19 these data follow predictable patterns reflecting local vegetation. We expect that insectivorous taxa occurring in more closed habitats will exhibit lower d13C values, reflecting the higher proportion of C3 resources present in such habitats. Conversely, those occurring in more open

13 habitats are expected to exhibit higher d C values, reflecting a greater availability of C4 resources. In addition, we conducted a pilot stable isotope study of fossil micromammals from the hominin-bearing sites Gladysvale and Sterkfontein to explore the feasibility of using such data to investigate past habitats.

MATERIALS AND METHODS

2.2.1 Field Collection Methods

This study was conducted in the Cradle Nature Reserve, located in the Cradle of

Humankind World Heritage Site, South Africa (see Figure 2.1). The reserve is located in the

Sterkfontein Valley, approximately 50 km northwest of Johannesburg, Gauteng Province at an elevation of about 1550m. The Sterkfontein Valley is well known for its Plio-Pleistocene aged hominin-bearing fossil cave deposits.

20

South Africa Lesotho

LM TR PV Gladysvale

Cradle of Humankind

Sterkfontein

Figure 2.1: Location of modern and fossil sampling sites in the Cradle of Humankind World Heritage Site, South Africa. Map Data: Google, DigitalGlobe.

The reserve lies within the Mesic Highveld Grassland vegetation type of the Southern

African Grassland Biome (Acocks, 1988; Mucina and Rutherford, 2006) and includes a mix of high inland plateau grassland and low inland plateau “bushveld” vegetation types. The landscape is a heterogeneous mix of open grasslands, wooded areas, and vleis (marshes). The vegetation is regularly exposed to fires and winter frosts. Annual temperatures range from -12° C to +39° C with a marked wet summer rainfall period (October to March) and a dry winter period (April to

September). Mean annual rainfall (1990-2009) is approximately 450mm (Climate Research Unit,

University of East Anglia; GPS 26.02 S, 27.88 E).

A significant sampling effort was undertaken within the reserve (12 separate microhabitats sampled, ~ 30,000 total trap nights for the dry season, where one trap night is 21 equivalent to one trap set for a 24 hour period). Shrews were trapped at three of the 12 microhabitats. These sites varied from open grassland to riverine woodland and yielded enough insectivores for hypothesis testing. Pieter’s Vlei (PV) represents the most open site and consists of a spring-fed marsh completely surrounded by grassland. Little Marsh (LM) is also a spring- fed marsh but is located within a mixed woodland environment. Tick River (TR) is a riverine woodland and represents the most closed microhabitat sampled. Characteristics of local vegetation were evaluated using step point line transects at each site. Transects were 250m in length and vegetation was recorded at 5m intervals along each transect. A 1x1m grid was placed on the ground every 5m and percent vegetative cover (including percent overall canopy cover, trees, forbs, grasses, sedges/reeds, and succulents) and ground cover (including proportions of plant cover, bare ground, rock, and detritus) were estimated visually. Additionally, all plant species within a 3m radius of each step point were identified (Mentis, 1981) (see Table 2.1).

Table 2.1: Characteristics of the three habitats sampled including vegetation type (in percentages), ground cover, and GPS coordinates for each site.

Vegetation Type (%) Ground Cover GPS Coordinates

Sedges/ Canopy Plant Bare Transect Grasses Reeds Forbs Trees Cover Cover Ground Rock Detritus Lat: 25°55’12” S Pieter’s Vlei 55 30 10 5 5 90 10 0 0 Lon: 27°52’10” E

Lat: 25°54’50” S Little Marsh 45 20 15 20 30 70 15 15 0 Lon: 27°50’59” E

Lat: 25°55’02” S Tick River 30 5 25 40 55 85 5 0 10 Lon: 27°52’39” E

Insectivores were trapped with the permissions of the Cradle of Humankind World

Heritage Site Management Authority, the Gauteng Provincial Government and the Cradle Nature

Reserve during May 2011 in collaboration the National Museum, Bloemfontein, South Africa.

Trapping was timed to correspond with the first seasonal frost to optimize trapping success rates 22 following the methods of Avenant (2011). Both ventilated Sherman-type metal box traps and wooden snap traps were used to sample specimens. Fifty of each trap type were used with a 5m spacing between traps along two 250m transects at each site. These 100 traps per site were in place for a 19-day trap period, totaling an early-dry season sampling effort of 5,700 trap nights at the three sites.

Each trap was baited with a mixture of peanut butter, oats, sunflower oil, and a yeast- based spread (Bovril®). Traps were checked, re-baited, and re-set at dawn and dusk each day within the sampling period. Upon capture, species identifications were made and basic biological data, including body mass, were recorded (Avenant, 2011). A small amount of hair was clipped from each specimen for isotopic analyses and these markings were used to ensure that the same individuals were not re-sampled within transects. For specimens sampled using snap traps, species identifications were independently confirmed using cranial morphology after processing and specimens were accessioned to the Bloemfontein National Museum Mammal Collection.

2.2.2 Stable Isotope Analyses

In total, hair samples from three soricid taxa and 143 individuals were collected for carbon and nitrogen isotope analyses. General ecological information for each taxon (Crocidura cyanea, Crocidura mariquensis, and Myosorex varius) is presented in Table 2.2. All three taxa are insectivorous, feeding predominantly on terrestrial, soft-bodied invertebrate prey. They differ primarily in body mass, and to a lesser degree in habitat preference, but are sympatric in all habitats sampled in this study (Skinner and Chimimba, 2005; Happold and Happold, 2013).

23 Table 2.2: Ecological information for each of the three species (C. cyanea, C. mariquensis, and M. varius) included in this study (data from Skinner and Chimimba, 2005; Happold, 2013).

Common Body Activity Locomotor Species Name Mass (g) Pattern Pattern Diet Habitat Preference

Crocidura Reddish- 5-11 g Active Terrestrial Insectivorous and Wide tolerance – has been cyanea grey Musk intermittently opportunistic, recorded in moist, dense Shrew both at night occasionally grassy habitats, stream and during the consuming borders, wet vleis, dry day vertebrate remains. bushveld, montane forest, karroid scrub, and fynbos. Often found amongst rocks, can tolerate relatively arid conditions (<500mm rainfall) and high altitudes.

Crocidura Swamp 4-20 g Active Terrestrial Insectivorous Occurs only in moist habitats mariquensis Musk intermittently (Possible including reed beds, swamps, Shrew both at night semi-aquatic and thick grass along river and during the adaptations) banks. day

Myosorex Forest 7-19 g Predominantly Terrestrial Insectivorous and Moist, dense grassland, varius Shrew nocturnal, but highly especially on the banks of intermittently opportunistic. streams, rocky areas, forests. active Occasionally throughout the cannibalistic and day especially will scavenge in midwinter. rodent carcasses.

Hair was first cleaned with ethanol and approximately 1 mg of sample was weighed into

Costech 3.5 X 5 mm pressed tin capsules. Samples were analyzed for 13C/12C and 15N/14N isotope ratios at the University of California Davis, using a PDZ Europa ANCA-GSL elemental analyzer interfaced to a PDZ Europa 20-20 isotope mass spectrometer. Samples were combusted at 1000° C in a reactor packed with chromium oxide and silvered copper oxide. Following combustion, oxides were removed in a reduction reactor (reduced copper at 650° C), and the resultant CO2 and N2 gases were separated using a Carbosieve GC column (65° C, 65mL/min) before entering the isotope ratio mass spectrometer. Stable isotope ratios are presented in delta

(d) notation in parts per thousand (‰) relative to the VPDB (Vienna PeeDee Belemnite) standard 24 for carbon and AIR standard for nitrogen. Analytical precision, based on repeated measurements of laboratory standards (approximately one of each standard for every ten samples), was less than 0.2‰ and 0.3‰ for d13C and d15N, respectively. Only data for d13C values are discussed here, but all d13C, d15N, and body mass data are available on Figshare

(http://figshare.com/s/49f14ce0897c11e594a006ec4b8d1f61)

Data were tested for normality and measures of central tendency are reported for habitats, taxa, and body mass. Median values, IQR and n are listed in parentheses and can also be found in

Tables 3 and 4. The effects of habitat, taxon, and body mass on isotopic composition, as well as the interactions between these variables were analyzed using nested ANOVA and 2-way

ANOVA (least squares method) in JMP Pro12 statistical software (SAS Institute Inc., 2015).

Omega-squared (ω2) effect sizes were calculated in MS-Excel.

RESULTS

Insectivore d13C values vary from a minimum value of -23.6‰ to a maximum value of -

11.4‰. d13C values are highest at PV (-14.4‰, -15.7 to -13.4‰, n=104), lowest at TR (-20.3‰,

-21.4 to -19.9‰, n=11), and intermediate at LM (median = -17.9‰, -18.5 to -16.7‰, n=28)

(Figure 2.2 and Table 2.3). 25

-11 -12 -13 -14

-15 -16 -17

(‰, VPDB) -18 hair

C -19 13 δ -20 -21 -22 -23 -24 Pieter’s Vlei Little Marsh Tick River

Site 13 Figure 2.2: Box plot of insectivore d Chair values by site.

Table 2.3: Stable carbon isotope compositions for each site sampled. δ13C values reflect the entire small mammal insectivore community sampled at each site.

13 δ C (‰)

Habitat Median IQR n

Pieter’s Vlei -14.4‰ -15.7 to -13.4‰ 104

Little Marsh -17.9‰ -18.5 to -16.7‰ 28

Tick River -20.3‰ -21.4 to -19.9‰ 11

26 Taxa differ in mean body mass. Crocidura cyanea is the smallest taxon (mean = 7.2g ±

SD 1.0, n=13), Myosorex varius (mean = 10.8g ± SD 1.5, n=76) is the largest taxon and

Crocidura mariquensis is intermediate between the two (mean = 8.7g ± SD 1.4, n=49) and

13 (Figure 2.3). There is no correlation between d Chair and body mass (see Figure 2.4).

15

14

13

12

11

10

9 Body Mass (g) 8

7

6

5

Crocidura Crocidura Myosorex C.cyanea cyanea C. mariquensis M. varius varius

Taxon

Figure 2.3: Box plot of body mass by taxon.

27

Site Little Marsh -12.0 Site Pieter’s Vlei Little Marsh Tick River -12.0 Pieter’s Vlei

-16.0 Tick River cyanea

-20.0 -16.0 cyanea

-24.0 R2 (TR) = 0.03 C. cyanea

-20.0 -12.0 mariquensis -16.0 -24.0 (‰, VPDB)

hair -20.0

-12.0 C 13 R2 (LM) = 0.10 δ -24.0 R2 (PV) = 0.00 C. mariquensis mariquensis -16.0 -12.0

-16.0 varius -20.0 -20.0 R2 (LM) = 0.12 2 -24.0 -24.0 R (PV) = 0.00 M. varius 5 6 7 8 9 10 11 12 13 14 15 -12.0 Body Mass (g)

13

-16.0 Figure 2.4: Regressions of d Chair against body mass by site and taxon. There isvarius no correlation between 13 d Chair and body mass.

-20.0 Taxa also differ in median d13C values although there is overlap between species (see

-24.0 Figure 2.4 and Table 2.4). Crocidura cyanea has the lowest median d13C value (-20.3‰, -21.5 to 5 6 7 8 9 10 11 12 13 14 15 -19.9‰, n=14), followed by Crocidura mariquensis (-16.0‰, -18.1 to -14.5‰, n=51) and

Myosorex varius (-14.1‰, -15.6 to -13.4‰, n=77).

28

Table 2.4: Stable carbon isotope compositions of each species and for each species by site.

13 δ C (‰)

Taxon Habitat n Median IQR

C. cyanea All 14 -20.3‰ -21.5 to -19.9‰

Pieter’s Vlei 2 -18.1‰ -- Little Marsh 2 -21.7‰ -- Tick River 10 -20.5‰ -21.5‰ to -20.1‰

C. mariquensis All 51 -16.0‰ -18.1 to -14.5‰

Pieter’s Vlei 37 -15.5‰ -16.7 to -14.2‰

Little Marsh 13 -18.1‰ -18.6 to -16.8‰ Tick River 1 -18.9‰ --

M. varius All 77 -14.2‰ -15.6 to -13.4‰

Pieter’s Vlei 64 -13.9‰ -15.0 to -13.2‰ Little Marsh 13 -17.1‰ -18.1 to -16.5‰ Tick River ------

A nested ANOVA was used to evaluate the influence of habitat, taxon, and body mass on d13C values. Species were nested within site and body mass was included as a co-variable

(Figure 2.4). The whole-model ANOVA is significant (F=32.1620, p< 0.0001) and effect tests show that site (F=20.2958, p<0.0001) and species nested within site (F = 8.3416, p<0.0001) are both significant, while body mass has no effect upon d13C values (F=0.3238, p=0.5703). Post- hoc analyses (Tukey’s HSD) imply that PV and LM do not differ and that species do not differ significantly between sites in several instances. However, the lack of difference between sites is driven by small sample sizes for some comparisons. When the underrepresented taxa at each site

(five specimens total; two C. cyanea each at PV and LM and one C. mariquensis at TR; Table

2.4) were excluded from analyses, post-hoc tests indicate that all sites clearly differ from one another. 29 Taxa are distributed differently across the three habitats. In equal trapping effort, C.

cyanea was the only species found in relatively high numbers at TR; low numbers of this species

were found at the more open sites PV and LM (see Table 2.4 for trapping frequencies). C.

mariquensis and M. varius occur in sufficient numbers for comparison at two sites, the more

open PV and the intermediately closed LM. To further explore the interaction between habitat

and taxon we ran a two-way ANOVA (least squares method) and included interaction terms.

Both site (F= 53.8436, <0.0001) and species (F= 10.1518, p=0.018) were found to be significant,

however, there was no significant interaction between site and species (F=1.5600, p=0.2140).

(See Figure 2.5 for Least Squares Means Plot). LS Means Plot

-12 LittleLittle Marsh Marsh Pieter’sPieter’s Vlei Vlei -14

(‰, VPDB) -16 hair C 13

δ -18

-20 C. mariquensis M. varius

Taxon

Figure 2.5: Least Squares Means Plot showing the lack of interaction between site and species.

After removing all non-significant interactions, both site (F=54.8845, p<0.0001) and

species (F=22.2934, p<0.0001) remain highly significant. Effect sizes (ω2) were calculated to

determine the degree to which habitat and taxon influence the variance in d13C values. Habitat

accounts for a greater percentage of the observed variance (habitat: ω2= 0.25; taxon: ω2= 0.10). 30 DISCUSSION

This study examined the carbon isotope values of three sympatric insectivorous micromammalian species from three different habitats in a southern African savanna.

In all comparisons, insectivore d13C values differ in the manner expected (i.e. habitats

13 with a greater proportion of C4 vegetation have higher d C values). The relatively high median

13 d C value characterizing insectivores at PV reflects the high proportion of C4 plant biomass surrounding the vlei. TR, the riverine woodland habitat, has the lowest median d13C value, the greatest degree of tree cover, and the highest proportion of C3 plant biomass. Insectivores at LM, characterized by mixed woodland, exhibit an intermediate median d13C value.

d13C values also differ taxonomically, but this is at least partly because taxa are differentially distributed across the three habitat types. Myosorex varius and Crocidura mariquensis occur in roughly equal numbers at PV and LM, but are rare at TR. In contrast,

Crocidura cyanea is most common at TR and poorly represented or absent at the other two sites.

This unequal distribution of taxa across sites is not unexpected given the different microhabitat preferences reported for each species (see Table 2.2). Regardless, median d13C values within individual taxa also differ in the manner expected across habitat types (see Figure 2.6). 31

cyanea mariquensis varius -11.0 C. cyanea C. mariquensis M. varius

-12.0

-13.0

-14.0

-15.0

-16.0

-17.0

(‰, VPDB) -18.0 hair

C -19.0 13 δ -20.0

-21.0

-22.0

-23.0

-24.0

-25.0 PV LM TR PV LM TR PV LM TR Site

13 Figure 2.6: d Chair values by species and site showing the directional shift in C. mariquensis and M. varius 13 d Chair values from site to site.

Having established that insectivore d13C values reflect habitat at a general level in our study area, we can now begin to investigate the degree to which insectivore values reflect the actual proportions of C3 and C4 vegetation in these habitats. For this study four metrics, percent tree cover, percent canopy cover, percent trees + forbs, and percent canopy cover + forbs were evaluated as proxies for total C3 vegetation. Forbs were included in two of our metrics in an effort to more accurately approximate the total percentage of C3 vegetation at a scale appropriate for small fauna. Though the inclusion of forbs is uncommon in studies focused on larger animals, forbs represent a legitimate food source for small mammals and insects and may contribute significantly to the carbon isotope signal (Scholtz and Holm, 1985; Skinners and Chimimba,

2005). To simulate an idealized micromammalian “ecological integrator”, we used a linear 32 mixing model with end member values for C4 and C3 vegetation of -12.4‰ and -27.0‰ respectively (data subset from Codron et al., 2013). A diet to hair fractionation of +1.0‰ for a small mammalian faunivore (see DeNiro and Epstein, 1978; Tieszen et al., 1983; Sponheimer et al., 2003b; Sare et al., 2005; Fox-Dobbs et al., 2007; Miller et al., 2008) was then applied resulting in an approximate C4 endmember of -11.4‰ and C3 endmember of -26.0‰. We generated regressions using vegetation proxy data (% tree cover, % canopy cover, % trees + forbs, % canopy cover + forbs) and insectivore mean d13C values for each transect (Figure 2.7).

33

-11.0 Legend Ideal Integrator Ideal Integrator Canopy Cover + Forbs -12.0 Trees + Forbs Canopy Cover -13.0 Trees d13C Mean -14.0 PV observed mean -15.0 -16.0

-17.0 LM observed mean -18.0 -19.0 C (‰, VPDB)

13 -20.0 δ TR observed mean -21.0 T CC T+F CC+F -22.0 Actual % C3 -23.0 vegeta.on present at site -24.0 Percentage C3 vegeta.on -25.0 predicted using observed insec.vore carbon value -26.0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

Percentage C3 Vegetation

13 Figure 2.7: δ C values and percentage C3 vegetation (where 100% C3 = 0% C4). The dashed line that crosses the graph at a diagonal represents an idealized micromammalian “ecological integrator” (Ideal Integrator). A linear mixing model using known regional end member values for C4 and C3 vegetation plus a diet to hair fractionation of +1.0‰ was used to generate a regression for an idealized integrator. Regressions were generated for each C3 vegetation proxy (T=Trees, CC=Canopy Cover, T+F=Trees + Forbs, and CC+F=Canopy Cover + Forbs) using recorded data and insectivore mean δ13C values for each habitat. δ13C means for each site are indicated on the graph.

Insectivore mean d13C values are, on average, slightly lower than would be predicted if insectivores were acting as perfect ecological integrators given known vegetation data for each transect (see Figure 2.7). When either tree cover or canopy cover alone are used as proxies, insectivore carbon values slightly overestimate the proportion of C3 vegetation and underestimate the proportion of C4 vegetation. These results are not necessarily unexpected given the close proximity of all transects to water (where C3 grasses often grow (Ellis et al., 1980)), and that shrews tend to forage under dense vegetation where they are concealed from predators and 34 where invertebrate prey are plentiful (Dickman, 1988, 1991; Churchfield, 1990; Skinner and

Chimimba, 2005). Intriguingly, canopy cover + forbs and tree cover + forbs yielded vegetation estimates that were much closer to those predicted if insectivores are acting as ideal ecological integrators in each habitat (see Figure 2.7). It is therefore likely that tree cover and canopy cover alone are somewhat poor indicators of available C3 resources at the scale relevant to small fauna such as insects and shrews.

Percent tree cover most consistently reflects the magnitude of change in vegetation (% difference) from site to site (e.g. the slope of the tree cover regression is similar to that of our idealized model). However, this measure is furthest from the proportion C3 predicted by an ideal integrator using observed d13C means for each site (see Figure 2.7). Alternatively, percentage

13 trees + forbs yielded estimates of C3 vegetation which were closest to those predicted given d C values for each transect (see Figure 2.7), but the magnitude of change from site to site was not as consistent (e.g. the slope is dissimilar from that of the model) as it is for trees alone. In any case, d13C values always shift in the direction expected regardless of which proxy is used to estimate

C3 vegetation.

The fact that insectivores (individual species and also the insectivore community as a whole) clearly evince a habitat signal is promising, both for modern ecological studies and for application to the fossil record. Small mammals are plentiful in many fossil deposits and insectivorous taxa occur frequently (Andrews, 1990; Avery, 2001; Reed, 2007).

As part of a related project, tooth enamel from modern and fossil micromammalian insectivores (including members of both Soricidae and Macrosceledidae) from the Cradle of

Humankind World Heritage Site, was sampled to determine its carbon isotope composition using laser ablation (Cerling and Sharp, 1996; Sharp and Cerling, 1996; Passey and Cerling, 2006). 35 Initial results from Sterkfontein (~2.4 Ma) and Gladysvale (~700 Ka) indicate that the fossils have significantly higher d13C values (p<0.0001) than modern specimens (after adjusting for the fossil fuel effect) (See Figure 2.8).

0.0 -1.0 -2.0 -3.0 -4.0

-5.0 -6.0 -7.0

(‰, VPDB) -8.0

enamel -9.0 C

13 -10.0 δ -11.0 -12.0 -13.0 -14.0 -15.0 Fossil Modern Age

13 Figure 2.8: δ Cenamel (VDPB) values from fossil and modern small mammal insectivores. For comparison, values have been adjusted from original laser method values to conventional acid method based upon Passey and Cerling (2006). Fossil values were adjusted by +0.5‰ (ε*laser-conv). Modern values were adjusted by 13 +0.3‰ (ε*laser-conv) and +1.5‰ (fossil fuel effect). Fossil δ Cenamel values are notably higher than modern enamel values.

This demonstrates a greater contribution of C4 vegetation to the dietary signal in the fossil taxa than in modern taxa from the same area. It is important to note, however, that all fossil specimens analyzed so far belong to Macroscelididae and all modern specimens belong to

Soricidae. Though both taxa are highly insectivorous and present in modern and fossil contexts, they differ somewhat both in their habitat tolerances and dietary behaviors (Skinner and 36 Chimimba, 2005). More data are needed to properly interpret carbon isotope values from fossil insectivores, but this preliminary look at available data from the Plio-Pleistocene of southern

Africa suggests that further investigation is warranted. We also acknowledge that inferring habitat data from fossil specimens requires close attention to taphonomy and other sources of potential bias, but which are beyond the scope of the present study (Andrews, 1990; Reed et al.,

2013).

CONCLUSION

The purpose of this study was to determine if the carbon isotope compositions of small mammal insectivores in a southern African savanna vary with habitat in a predictable manner.

We determined that insectivore carbon isotope compositions track changes in the percentages of

C3 and C4 vegetation in the local environment.

Future studies across a broader array of habitats, and which are explicit with regard to seasonality, taxonomy, and relative abundance, are needed to confirm the link between insectivores and environment. If properly developed, insectivore carbon isotope analyses may prove a valuable tool for resolving questions about ancient vegetation composition and structure at a finer scale than is possible with larger fauna. In so doing, it could contribute to our understanding of hominin landscape use and the impact of environmental change on hominin evolution. 37 3 STABLE CARBON ISOTOPE ECOLOGY OF SMALL MAMMAL TOOTH ENAMEL

FROM BARN OWL ROOSTS IN THE STERKFONTEIN VALLEY

INTRODUCTION

Isotopic analysis of fossil tooth enamel is frequently employed in studies of large mammals as a means to understand dietary ecology, habitat preference, and to reconstruct past environments (e.g., Koch et al., 1994; Cerling et al., 1997; Luyt and Lee-Thorp, 2003;

Schoeninger et al., 2003; Sponheimer and Lee-Thorp, 2003; Kingston and Harrison, 2007; Lee-

Thorp et al., 2007; Levin et al., 2008; Bedaso et al., 2010, 2013). Stable isotope analyses of small mammal faunas, however, are less common, largely due to the methodological difficulties inherent in sampling very small teeth and a poor understanding of small mammal isotopic ecology. Nevertheless, isotopic analyses of small mammal fossil assemblages hold promise for a variety of reasons. For one, small mammals occur in many archaeological and paleontological sites and when present they are often abundant making it easy to obtain a statistically robust sample. In addition, small mammals are diverse in habitat preference and dietary habits. They occur in habitats ranging from desert to forest and everything in between (Walker, 1999). Yet, unlike larger fauna, they have quite limited lifespans and home ranges and are therefore both temporally and spatially constrained. These constraints mean that they have potential for fine- scale reconstruction of past habitats. Whether or not such fine-scale reconstruction can be realized, small mammal isotopic datasets are important complements to those derived from larger fauna, which currently dominate the literature.

The use of small mammals in paleontological reconstructions has usually been restricted to more traditional faunal analytical methods reliant upon the association of modern taxa with preferred habitats (Cartmill, 1967; Jaeger, 1976; Wesselman, 1984, 1985, 1995; Fernandez- 38 Jalvo, 1998; Denys, 1999; Avery, 2001; Reed, 2003, 2007). However, researchers have begun to explore small mammal stable isotope ecology in both modern and paleontological contexts with promising results (see Navarro et al., 2004; Hopley et al., 2006; Yeakel et al., 2007; Heran et al.,

2010; Gehler et al., 2012; Hynek et al., 2012; Kimura et al., 2013; Jeffrey et al., 2015 Smiley et al., 2015; Commendador and Finney, 2016; Patterson et al., 2016). Recent work suggests that the isotopic compositions of small mammals have potential for recording dietary behavior, habitat composition, and environmental change over short and long timescales (Gehler et al.,

2012; Hynek et al., 2012; Robb et al., 2012; Dammhahn et al., 2013; Kimura et al., 2013; Symes et al., 2013; Codron et al., 2015; Jeffrey et al., 2015; Smiley et al., 2015; Commendador and

Finney, 2016). For instance, studies of fossil microfauna from Africa document the presence of

13 C4 vegetation at hominin-bearing sites and suggest a pattern of increasing mean d C values through time (Hopley et al., 2006; Yeakel et al., 2007; Patterson et al., 2016; but see Thackeray et al., 2003).

Currently, there are more carbon isotopic studies of fossil tooth enamel than there are studies exploring small mammal carbon isotope ecology in modern ecosystems. This is problematic because, without a clear understanding of modern small mammal isotope ecology, it can be difficult to extrapolate to the past. Moreover, studies vary widely in terms of methods used for sample preparation (e.g. conventional acid preparation vs. laser ablation), material type

(enamel, dentin, bone, whole tooth), tooth type (incisor or molar), and sampled taxonomic diversity. These methodological differences can significantly influence patterns in isotopic datasets. Bone and dentin, for example, are known to be more susceptible to diagenetic change than enamel. Similarly, for rodents, tooth type sampled matters because molars and ever-growing incisors can reflect markedly different dietary periods of the animal’s life. Though some studies 39 have begun systematically addressing these concerns (see Gehler et al., 2012; Hynek et al., 2012;

Jeffrey et al., 2015; Smiley et al., 2015) large gaps remain in our knowledge of small mammal isotopic ecology across spatial scales, seasons, and heterogeneous landscapes. Moreover, few studies have been conducted in African ecosystems, where C4 plants are abundant and many landscapes relevant to hominin evolution persist. This paucity of data is further compounded in the context of the fossil record, where the influences of accumulation processes (taphonomy) must also be considered in interpreting fossil isotope values.

In an effort to begin addressing these gaps, we examine the stable isotope ecology of modern and fossil small mammals from owl roosts in the Sterkfontein Valley, South Africa. Our explicit goals were 1) to examine variation in small mammal carbon isotope compositions within taxa, between taxa, and between tooth types, and 2) to evaluate how well small mammal isotope data reflect vegetation composition local to owl roosts. We chose to use owl pellet accumulations for this study because barn owl diets show high fidelity to the small mammal community in their hunting range (usually ~ 1-2 km) (Andrews, 1990; Avenant, 2005; Reed,

2007; Terry, 2010) and because barn owls have been identified as the primary accumulating agent of microfauna at fossil sites in the area (Avery 2001).

We anticipate high variation in δ13C values within taxa and expect differences between taxa with specific dietary habits as well as differences between tooth types (i.e. ever-growing incisors vs. permanent molars). We expect that small mammal community isotopic compositions will reflect the vegetation local to roosts. Small mammals from roosts in more closed (i.e. wooded) habitats should have lower δ13C values than those from roosts in more open environments. In addition, we explore how well the micromammal carbon isotopic compositions correspond to the known vegetation around each roost after taxonomic filters and adjustments for 40 relative abundance are applied to the data. Finally, we present a comparison of fossil micromammal stable isotope data across three Plio-Pleistocene hominin-bearing sites in the

Sterkfontein Valley to determine how these and similar datasets may be used to help clarify patterns of vegetation change and composition relevant to hominin evolution.

MATERIALS AND METHODS

The field component (extant small mammal investigation) of this study was conducted in the John Nash and Malapa Nature Reserves, located in the Cradle of Humankind World Heritage

Site, South Africa (Figure 3.1). The reserves are located in the Sterkfontein Valley, approximately 50 km northwest of Johannesburg, Gauteng Province at an elevation of about

1450-1550 m. The Sterkfontein Valley is well known for its Plio-Pleistocene hominin-bearing fossil cave deposits. 41

Cradle of Humankind World Heritage Site

South Africa Lesotho

Gladysvale Malapa Roost Site Roost Site

Gladysvale Kimberley Africa Roost Site

Swartkrans

Sterkfontein

(-25.9254° S, 27.7674° E) Figure 3.1: Locations of Modern Roost sites (trees) and Fossil sites (filled circles) included in this study. Map Data: Google, DigitalGlobe

The reserve lies within the Mesic Highveld Grassland vegetation type of the Southern

African Grassland Biome (Acocks, 1988; Mucina and Rutherford, 2006) and includes a mix of high inland plateau grassland and low inland plateau “bushveld” vegetation types. The landscape is a heterogeneous mix of open grasslands, wooded areas, and vleis (marshes). The vegetation is regularly exposed to fires and winter frosts. Annual temperatures range from – 12° C to + 39° C with a marked wet summer rainfall period (October to March) and a dry winter period (April to

September). Mean temperature (1990-2009) in the warmest month (January) is 23.5° C and coolest month (July) is 11.9° C while mean annual rainfall is approximately 450 mm (GPS

26.02° S, 27.88° E) (Climate Research Unit, University of East Anglia).

Three modern barn owl roosts were selected from within the reserve, each from a different microenvironment. The Kimberley roost site (KRS) is located in an abandoned building surrounded by open grassland (for image of site see Figure 3.3A). The Malapa (MRS) and

Gladysvale (GRS) modern roosts are surrounded by significant tree cover and are similarly vegetated (Figure 3.3B, Figure 3.3C). Habitats around the roosts were characterized using step- point line transects at each site. Transects were 250 m in length and vegetation was recorded at 5 m intervals along each transect. A 1x1 m grid was placed on the ground every 5 m and percent vegetative cover (including percent overall canopy cover, trees, forbs, grasses, sedges/reeds, and succulents) and ground cover (including proportions of plant cover, bare ground, rock, and detritus) were estimated visually (see Table 3.1). Additionally, all plant species within a 3 m radius of each step point were identified (Mentis, 1981). 43

Table 3.1: Vegetation data for modern roost sites

Vegetation Type (%) C3 estimators (%) Canopy Cover (%) + Modern Roost Site Grass Sedges/Reeds Forbs Trees Canopy Cover (%) Forbs (%) 1 – % Grasses GPS coordinates* 25°54'54.82"S Kimberley (KRS) 95 0 5 5 5 10 5 27°49'4.18"E 25°53'50.91"S Malapa (MRS) 85 0 10 5 20 30 15 27°48'7.06"E 25°43'49.93"S Gladysvale (GRS) 65 0 5 5 30 35 35 27°46'22.96"E

* Coordinates are vegetation transects adjacent to roosts 44

Owl pellets, both new and old/disaggregated, were collected in June 2010 and 2011 from the Malapa, Gladysvale, and Kimberley modern roosts. Thus, our samples represent aggregations of micromammals collected by owls over multiple seasons, and in some cases, over multiple years. Pellets were mechanically processed using forceps, and micromammal fauna were identified to genus and species (where possible) using cranial and dental characteristics and the aid of a comparative collection at the Ditsong National Museum, Pretoria, SA (see Table 3.2 for ecological information on taxa). The relative abundance of each taxon was calculated based on the minimum number of individuals in each roost, as represented by cranial elements. 45 Table 3.2: Ecological data for small mammal taxa (Happold, 2013 and Skinner and Chimimba, 2005; Taxonomy after Skinner and Chimimba).

Taxon Common Name Activity Pattern Locomotor Pattern Diet Habitat Preference

Muridae Dendromus sp. African Climbing Mostly nocturnal Arboreal, adept at Primarily seeds, some insects and Long grasslands, bracken, dense Mice climbing slender berries scrub, grassy wetlands, grasses and shrubs, subalpine or alpine vegetation. semi-prehensile tail Mastomys sp. Multimammate Nocturnal Terrestrial Opportunistic omnivores Grasslands, wooded savanna, Mice cultivated areas, human settlements Micaelamys Namaqua Veld Rat Nocturnal Terrestrial and semi- Opportunistic, diet is primarily grass Savanna and semi-arid areas, namaquensis arboreal and foliage, some seeds, some often associated with rocky insects outcrops Mystromys African White- Nocturnal Terrestrial No data available from wild Grasslands, fynbos, montane albicaudatus Tailed Rat populations. Digestive system grasslands, karoo. suggests predominantly plant foods. Otomys sp. Vlei Rats Mainly crepuscular Terrestrial Grass Damp grasslands and swamps Steatomys sp. Fat Mice Nocturnal Terrestrial Predominantly granivorous. Also Grasslands, woodland savannas, consumes vegetable material and fields, edges of rivers and insects. swamps. Generally not found in forests and montane grasslands. Gerbillinae spp. Gerbils Nocturnal Terrestrial Predominantly granivorous but may Sandy soils and alluvium, in dry, eat a variety of foods including open environments including insects open grasslands, wooded savannas, scrub. Bathyergidae Cryptomys sp. Mole-Rats Irregular Fossorial Herbivorous, specialize on bulbs, Fynbos, grassland, savanna, and corms and tubers. karoo. Occurs in softer soils suitable for burrowing. Soricidae Shrews Mostly nocturnal Terrestrial/Semi- Insectivorous Varied, generally associated aquatic with water. Macroscelididae Mostly diurnal Terrestrial Predominantly insectivorous, some Arid lowlands, such as (Elephantulus/ Elephant vegetable matter depending on savannahs, scrublands, rocky Macroscelides) Shrews/Sengi’s species outcrops, and deserts 46 Samples of at least 5 individuals per taxon (where possible) were selected from each modern owl roost for stable isotope analysis (N = 110). For rodents, maxillary incisors were removed from the crania, and the outer surface (often covered with organic material) was removed using a Dremel tool outfitted with a diamond-encrusted burr. For a subset of individuals, molar teeth were also removed, and the surfaces were mechanically cleaned. For shrews, hemi-mandibles were cleaned mechanically. Enamel surfaces were then blasted with compressed N2 gas to remove particulates.

For the fossil component of this study, small mammal remains from three fossil sites,

Sterkfontein Member 4 (Sts) (~ 2.6 – 2.2 Ma) (Herries and Shaw, 2011), Swartkrans Member 1

(Sk) (~ 1.8 Ma) (Vrba, 1985), and Gladysvale Fossil Site (GV) (~ 700 Ka) (Hall et al., 2006) were analyzed. Depending on preservation, fossil specimens were identified to the most specific taxonomic level possible. Relative abundance was based upon MNI and includes both newly identified specimens (N = 391; this study) and previously published data for the sites (see Avery,

2001). At least 5 specimens from each of the most common taxa at three fossil sites (N = 76) were selected for isotopic analysis. Incisors were preferentially chosen for analysis because they are easier to sample (i.e. have a greater surface area), but paired incisors and molars were also selected from a small subset of the sample.

3.2.1 Isotopic Analyses

Laser ablation-gas chromatography-stable isotope ratio mass spectrometry (LA-GC-

IRMS) was used to measure δ13C and δ18O values in incisor and molar enamel (sensu Sharp and

Cerling, 1996; Passey and Cerling, 2006). Discussion of oxygen isotope data is beyond the scope of the current paper, but δ18O values are included for completeness. The laser system, housed at the Department of Earth and Planetary Science of Johns Hopkins University, 47 incorporates custom-built features developed to optimize the analysis of very small enamel samples. Samples are loaded onto a platform inside a glass chamber and subjected to helium flow to purge the chamber. Outgassing of CO2 from sample surfaces can take several hours or much longer for large porous samples (Passey and Cerling, 2006). Chambers were allowed to purge in helium for at least 6 hours.

The sample platform inside the laser chamber rotates to allow irregularly shaped samples

(such as curved rodent incisors) to be appropriately positioned under the laser, which itself moves on an x-y-z stage allowing for precise spatial control of the specimens beneath the laser.

A Photon Machines Fusions CO2 laser emits short bursts (0.01 s) of low power radiation

(~5 Watts) at a wavelength of 10.6 µm to ablate samples in the glass chamber. The laser was set to produce a line of partially overlapping spots (approximately ~ 200 µm in diameter) typically spaced ~100-150 µm from one another. CO2 gas liberated from several ablation events is carried by helium gas into a semi-automatic extraction line where potential contaminants (organics, water vapor) are removed cryogenically. The CO2 is then cryogenically focused, passed through a gas chromatography column (Poraplot Q, 25 m length, 0.32 m inner diameter, held at 70° C,

5.0 grade He carrier gas) and finally admitted into a Thermo 253 mass spectrometer via a ConFlo interface.

Up to four separate analyses were performed on each tooth. The lines of spots for each analysis were oriented in the apical-cervical direction, and these were averaged to produce single

δ13C and δ18O values for each tooth. Our spatial sampling strategy was to integrate as much of the enamel surface as possible to obtain an average value for each tooth.

Stable isotope compositions are expressed as d values, dx = 1000 (Rsample / Rstandard -1)

13 18 13 18 where x is C or O and Rsample and Rstandard are the molar ratios of the heavy ( C, O) to light 48 (12C,16O) isotope for the sample and the standard. d13C and δ18O are expressed in per mil (‰) relative to the international standard Vienna Pee Dee Belemnite (VPDB) and Vienna Standard

Mean Ocean Water (VSMOW) respectively. Data were normalized to the VPDB scale via

13 injections of an internal CO2 standard calibrated to the international standard NBS-19 (δ C =

18 1.95 ‰, δ O = –2.19 ‰). The CO2 standard is injected into the extraction line and treated in the same manner as sample CO2 generated with the laser. During analytical runs, in-house enamel standards were measured periodically to monitor for possible isotope effects related to laser ablation. Blank runs were performed periodically between sampling events to characterize the background signal and to monitor the behavior of the chamber through time. Sample yields were monitored to ensure that they exceeded the size of the background signal by at least tenfold.

Laser ablation and conventional phosphoric acid hydrolysis produce δ13C and δ18O values that have a 1:1 relationship. There is, however, an offset for both carbon and oxygen isotope values. Passey and Cerling (2006) calculated the laser-conventional isotope enrichment (ε*) to be –0.3 ± 1.1 ‰ for δ13C and –6.4 ± 0.7 ‰ for δ18O for modern enamel, and –0.5 ± 0.8 ‰ for

δ13C and –5.1 ± 1.2 ‰ for δ18O for fossil enamel. The larger offset in δ18O results from the mixing of the oxygen-bearing phases of enamel during laser ablation. The phosphoric acid method liberates CO2 from the carbonate fraction of enamel alone, which accounts for ~ 6 % of the oxygen in enamel apatite. Phosphate accounts for ~ 90 %, and hydroxide ~ 3 % (Elliot,

1997). Enamel phosphate δ18O is approximately 9 ‰ lower than carbonate δ18O (Bryant et al.,

1996; Iacumin et al., 1996). Modern enamel δ13C values were adjusted + 0.3 ‰ and fossil enamel δ13C values were adjusted + 0.5 ‰ to make the values comparable to conventional phosphoric acid methods (Passey and Cerling, 2006). 49 RESULTS AND DISCUSSION

3.3.1 Incisor-Molar Pairs

Our carbon isotope analyses of incisors and molars from the same individual (incisor- molar matched pairs, N = 14) reveal no consistent pattern or statistically significant differences between tooth types overall (Incisor = –9.8 ‰; Molar = –8.6 ‰; Mean difference = –1.2 ‰

± 1.1 ‰ (SE); t-ratio = –1.10; df = 13, 2-tailed p = 0.29). Values for both δ13C and δ18O are presented in Table 3.3.

Nevertheless, incisors and molars frequently differ from one another, differences can be quite large (> 10 ‰), and the difference can vary in either direction. Incisors range from a minimum value of –17.5 ‰ to a maximum value of –4.7 ‰ while the range in molar values is slightly less, from –14.8 ‰ to –3.9 ‰. The maximum difference between any matched incisor- molar pair is –11.3 ‰, though the average difference between matched pairs is –1.2 ‰. For most taxa, no consistent patterning was observed in the difference between incisors and molars (Figure

3.2). One species from the fossil sites, Mystromys albicaudatus, exhibits a statistically significant offset in incisor and molar pairs. For this taxon, incisors are on average 1.5 ‰ higher than molars

(N = 6, t-ratio = 2.50, df = 5, p = 0.0273). 50 Table 3.3: Carbon and oxygen isotope data for paired molars and incisors

Incisors Molars Difference (Incisor-Molar)

18 18 13 18 13 δ O ‰ 13 δ O ‰ Δδ C ‰ Δδ O ‰ δ C ‰ (PDB) δ C ‰ (PDB) Taxon Specimen No. Modern/Fossil Site (SMOW) (SMOW) (PDB) (SMOW) Gerbillinae SASM 28 Modern Gladysvale Roost -9.0 28.3 -3.9 25.8 -5.2 2.5 Gerbillinae SASM 31 Modern Gladysvale Roost -11.4 29.2 -7.4 28.3 -4.0 0.9 Mastomys coucha SASM 7 Modern Malapa Roost -17.5 23.6 -6.2 23.7 -11.3 -0.1 Mastomys coucha SASM 9 Modern Malapa Roost -11.7 27.3 -4.8 23.6 -6.8 3.6 Micaelamys namaquensis SASM 20 Modern Malapa Roost -15.0 28.9 -14.8 24.4 -0.2 4.5 Micaelamys namaquensis SASM 21 Modern Malapa Roost -14.1 32.5 -13.3 24.8 -0.9 7.7 Otomys irroratus SASM 1 Modern Malapa Roost -4.7 24.7 -7.2 28.4 2.5 -3.6 Otomys irroratus SASM 3 Modern Malapa Roost -6.1 22.0 -5.0 25.7 -1.1 -3.7 Mystromys albicaudatus SW 10 Fossil Swartkrans -6.3 26.2 -10.4 27.5 4.1 -1.2 Mystromys albicaudatus SW 11 Fossil Swartkrans -8.3 26.6 -9.9 27.3 1.6 -0.6 Mystromys albicaudatus SW 2 Fossil Swartkrans -9.9 23.5 -9.4 24.7 -0.5 -1.1 Mystromys albicaudatus SW 5 Fossil Swartkrans -6.4 32.3 -8.7 27.1 2.3 5.1 Mystromys albicaudatus SW 7 Fossil Swartkrans -8.2 27.1 -9.5 23.2 1.3 3.8 Mystromys albicaudatus SW 9 Fossil Swartkrans -8.7 24.6 -9.4 26.5 0.6 -1.9 51

Incisor -2 Molar -3

-4

-5

-6

-7

-8

-9

C ‰ (PDB) -10 13 δ -11

-12

-13

-14

-15

-16

-17

-18 SW 2 SW 5 SW 7 SW 9 SW 10 SW 11 SASM 7 SASM 9 SASM 1 SASM 3 SASM 28 SASM 31 SASM 20 SASM 21 Mystromys Gerbillinae Mastomys Micaelamys Otomys Swartkrans M1 Gladysvale Malapa Fossil Modern

Figure 3.2: d13C for incisor-molar matched pairs by site and taxon.

52 Inter-tooth variation in d13C values has been documented previously for a few small mammal species (Gehler et al., 2012; Jeffrey et al., 2015), though the average and maximum differences observed in this study exceed those reported in these previous studies. Inter-tooth variation in carbon isotope compositions observed here are not wholly unexpected given the known generalist dietary tendencies of species included in this study and differences in diet during tooth formation. Molars of the species included in this study are formed primarily in utero and during the first weeks after birth and therefore reflect a singular period in the animal’s life.

Thus, molars probably exhibit less variation in d13C than ever-growing incisors, which constantly change, often turning over completely in as little as 4 weeks. For a given taxon, low variation in molar δ13C values might be a reflection of a preferential breeding/weaning season. Similarly, consistency in incisor δ13C values may be indicative of consistency in diet or of a common time of death. The incisor-molar data for Mystromys are consistent with such a scenario. However, this particular species is currently regionally extinct, currently designated by the IUNC as vulnerable, and was not found in any of the modern assemblages we collected, making interpretation of the pattern difficult (Avenant et al., 2015). Regardless, we concur with the caution of Jeffrey et al. (2016) that tooth type should be carefully considered in study design when using an isotopic approach to paleontological research, especially in small-bodied animals with fast tissue growth and turnover. For the remaining analyses in this study, incisors (including incisors only for matched pairs) were used wherever possible to maintain consistency.

3.3.2 Modern Roost Site Comparisons

When carbon isotope ratio data for all taxa are pooled, sites do not differ significantly

(the means are within ~ 2 ‰ of one another). In general, d13C values of small mammals sampled in this study vary widely (~ 17 ‰ range for each site). Intra-taxon variation is also generally high 53 but patterns reflecting community structure and niche differentiation emerge when sites are further broken down by taxon (Figure 3.3). For example, Dendromus sp., a grass seed specialist associated with climax vegetation (Skinner and Chimimba, 2005; Avenant, 2011), has

13 consistently higher mean d C values than other taxa, reflecting the greater contribution of C4 resources to its diet (see Figure 3.3 and Table 3.4). Conversely, the dietary generalist

Micaelamys namaquensis and Mastomys coucha often have lower average d13C values across sites than most other taxa (see Figure 3.3 and Table 3.4) 54

Kimberley Roost Site Malapa Roost Site Gladysvale Roost Site

0 0 0 -2 -2 -2 -4 -4 -4 -6 -6 -6 -8 -8 -8 -10 -10 -10 C ‰ (PDB)

13 -12 -12 -12 δ -14 -14 -14 -16 -16 -16 -18 -18 -18 -20 -20 -20

Otomys Otomys Otomys Mastomys Mastomys Steatomys Mastomys Cryptomys Dendromus Micaelamys Micaelamys Dendromus Gerbillinae Micaelamys

Figure 3.3: Box plots d13C by roost and taxon with images of the vegetation at each modern roost site. 55 Table 3.4: Summary statistics for carbon and oxygen isotope data by taxon for each modern roost and fossil site (Mean±SD (N)).

Modern Fossil

Kimberley Malapa Gladysvale Sterkfontein M4 Swartkrans M1 Gladysvale Fossil

13 18 13 18 13 18 13 18 13 18 13 18 δ C ‰ δ O ‰ δ C ‰ δ O ‰ δ C ‰ δ O ‰ δ C ‰ δ O ‰ δ C ‰ δ O ‰ δ C ‰ δ O ‰ Taxon (PDB) (SMOW) (PDB) (SMOW) (PDB) (SMOW) (PDB) (SMOW) (PDB) (SMOW) (PDB) (SMOW)

Muridae Dendromus sp. -3.2 ± 3.1 (8) 24.9 ± 1.5 (8) -3.5 ± 1.2 (6) 25.8 ± 2.1 (6) Mastomys sp. -6.3 ± 2.5 (6) 23.8 ± 0.8 (6) -14.2 ± 3.6 (9) 23.2 ± 3.0 (9) -13.6 ± 3.8 (7) 24.7 ± 1.6 (7) Micaelamys namaquensis -10.9 ± 4.4 (8) 25.3 ± 1.5 (8) -15.0 ± 1.6 (6) 30.3 ± 2.5 (6) -14.9 ± 2.5 (8) 30.6 ± 4.1 (8) -5.3 (1) 31.1 (1) Mystromys albicaudatus -8.3 ± 1.7 (11) 27.6 ± 1.4 (11) -8.1 ± 1.4 (11) 27.4 ± 2.7 (11) -7.4 ± 1.9 (7) 27.8 ± 2.3 (7) Otomys sp. -10.4 ± 1.5 (8) 25.5 ± 1.6 (8) -7.0 ± 2.0 (5) 27.9 ± 4.9 (5) -7.7 ± 2.4 (10) 27.7 ± 3.0 (10) -4.4 ± 1.6 (11) 28.1 ± 1.9 (11) -10 ± 2.7 (6) 26.2 ± 1.4 (6) -6.0 ± 2.8 (6) 28.7 ± 3.9 (6) Steatomys sp. -4.5 ± 5.0 (5) 24.2 ± 2.4 (5) Murinae indet. spp. -11.3 ± 7.6 (2) 24.2 ± 0.7 (2) -6.3 ± 1.7 (7) 29.7 ± 1.8 (7) Gerbillinae indet. spp. -11.3 ± 1.4 (5) 25.9 ± 3.2 (5) Bathyergidae Cryptomys sp. -10.7 ± 3.8 (6) 23.6 ± 0.8 (6) -6.6 ± 3.3 (5) 25.2 ± 0.8 (5) Soricidae -11.7 ± 2.2 (6) 23.9 ± 1.5 (6) -11.1 ± 1.9 (5) 21.6 ± 3.1 (5) Macroscelididae -4.6 ± 1.5 (6) 25.7 ± 0.9 (6) -3.5 ± 1.7 (5) 27.9 ± 2.1 (5) 56

Figure 3.4: Box-and-whisker plots of δ13C values for each modern roost site including A) all measured taxa; B) members of the order Rodentia only; C) three dominant rodent taxa (Mastomys, Micaelamys, Otomys); D) two dominant generalist taxa (Mastomys, Micaelamys). In panels C and D, sites marked with an asterisk are significantly different from the others (p < 0.05). 57 To determine if discrimination between modern roost sites using mean differences in d13C is possible under different conditions, we analyzed the data in four different ways, utilizing different “exclusion criteria”. Box plots of these analyses are represented in Figure 3.4 A-D.

Figure 3.4A illustrates data with no exclusion criteria applied. All data for all sites are included in this panel. The site means are indistinguishable from one another (p = 0.2022). We therefore do not find support for our initial hypothesis that there will be an overall difference between habitats at the level of the small mammal community mean.

Figure 3.4B includes d13C values for rodents only (Order Rodentia) and excludes shrews

(Order Soricomorpha). This adjustment is warranted because 1) we do not have values for shrews at all sites, 2) shrews are exclusively insectivorous and therefore differ significantly from most rodents in their dietary behavior, and 3) rodent teeth are more readily identifiable to species, which allows for better taxonomic control. Under this criterion, no significant differences were found at the mean level (the means are within 2.4 ‰ of one another) (p=

0.1422). Again, these results are not unexpected given the high degree of variation in d13C values across the diversity of rodent taxa included here.

Figure 3.4C includes d13C values for only a subset of taxa represented at all roosts. Three taxa, Mastomys sp., Michaelamys sp., and Otomys sp., are the most common rodent taxa at all three sites. Using this criterion, we found significant differences between sites (Tukey’s HSD, P

< 0.05). In pairwise comparisons mean d13C is higher at Kimberley (KRS) than both of our other roost sites, but this difference is only significant relative to Malapa (MRS) (p = 0.0388). The higher mean value observed at Kimberley (KRS) ( = –9.5 ‰) reflects the greater proportion of

C4 vegetation present at this site compared to Malapa (MRS) ( = –12.6 ‰) and Gladysvale

(GRS) ( = –11.6 ‰) and is expected based upon the vegetation data collected for these sites. 58 Finally, Figure 3.4D, we include only the dominant taxa that are dietary generalists

(Mastomys sp. and Micaelamys sp.). Otomys sp. has been excluded here because it is understood to be a strict herbivore and generally confined to marshy vlei habitats. As dietary generalists,

Mastomys sp. and Michaelamys sp. are not only expected to integrate resources in a more unbiased manner than the dietary specialist Otomys sp. but they also occur in a wider variety of microhabitats around each roost site. Site means differ significantly in the direction expected under these criteria (Tukey’s HSD, P < 0.01). Pairwise comparisons reveal that mean d13C value at Kimberley (KRS) ( = –9.0 ‰) is significantly higher than both Malapa (MRS) ( = –14.5

‰, p < 0.0003) and Gladysvale (GRS) ( = –14.3 ‰, p < 0.0005), while the latter two sites are indistinguishable.

3.3.3 Relative Abundance for Modern Roost Sites

To approximate a balanced sample design, an equal number of replicates were chosen from each taxon whenever possible. However, taxa are not equally distributed in nature – certain common taxa usually dominate a community while others are rarer. It is useful in ecological studies to account for this discrepancy in the relative abundance of taxa when interpreting data.

Therefore, we also used the minimum number of individuals to estimate the relative abundance of each taxon at each modern roost site and weighted mean d13C values for each site accordingly.

To evaluate the effect of adjusting mean d13C values for relative abundance in predicting known vegetation characteristics, we first developed a theoretical small mammal community

13 mean d C value for each roost site. Percentages for canopy cover, trees, forbs, and grasses were recorded for each roost site and we estimated percentage C3 vegetation by summing percentage canopy cover and percentage forbs. While canopy cover alone is more commonly used as a primary proxy for C3 vegetation, this measure is likely more appropriate at the scale of a large 59 mammal. At the finer scale of small mammals, landscapes can be quite heterogeneous and even patchily distributed forbs likely contribute significantly to diet (Skinner and Chimimba, 2005;

Happold, 2013).

13 To generate theoretical small mammal community mean d C values we used a linear mixing model with end member values for C4 and C3 vegetation of –12.4 ‰ and –27.0 ‰ respectively (data subset from Codron et al., 2013). A diet to enamel fractionation of +10.0 ‰ for rodents (see DeNiro and Epstein, 1978) was then applied, resulting in an approximate C4

13 endmember of –2.4 ‰ and C3 endmember of –16.0 ‰. The d C values for each site are intended to reflect a “theoretical” isotopic mean that would result if each small mammal community perfectly integrated the carbon composition of the vegetation surrounding the roost site.

Theoretical community means used to generate the linear mixing model are plotted as points on the solid line depicted in Figure 3.5A-D. Both unadjusted mean values (unadjusted for relative abundance) and adjusted mean values (adjusted for relative abundance) are plotted in each panel for evaluation against the linear model. In particular, we evaluated the average difference of each measure from the linear model, how well the magnitude of change from site to site was preserved (slope), and how much points varied around each regression line (r-squared).

We applied each of the permutations (exclusion criteria) used previously (all taxa, rodents only, dominant rodent taxa, dominant generalist rodent taxa) (see Figure 3.4).

In all cases, small mammals appear to overestimate actual proportions of C3 vegetation in their immediate habitats. Unadjusted and adjusted d13C values are consistently lower than the theoretical community means for each site. These results imply that small mammals have a dietary bias towards C3 foods relative to the measured proportion of C3 vegetation surrounding each roost, or that the proxies commonly used to estimate C3 vegetation are too coarse at this 60 scale of analysis. Both are likely true. This is not without precedent as an abstract by Fox-Dobbs et al. (2011) reports that Kenyan small mammal diets skew towards C3 resources even in strongly

C4 grassland dominated habitats.

Adjusting means for relative abundance does not always improve slope (i.e. produce slopes similar to that of the theoretical model). Using rodents only (roughly equal sample sizes of rodents, excluding shrews) adjusted for relative abundance (Figure 3.5B) produces our best predictive measure in these analyses. The slope of the regression differs the least from the slope of the theoretical linear model and r-squared is high (0.99). This indicates that observed means

(adjusted for relative abundance) preserve the magnitude of change between sites with high fidelity. 61

-2 -2 Theoretical Values -4 -4 Unadjusted Mean -6 -6 Mean Adjusted for Relative Abundance -8 -8

-10 -10

-12 -12

-14 -14

-16 -16

0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 -2 -2 C ‰ (PDB) 13

δ -4 -4

-6 -6

-8 -8

-10 -10

-12 -12

-14 -14

-16 -16

0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 % C3 Vegetation Figure 3.5: Theoretical and observed means (both unadjusted and adjusted for relative abundance) for each modern roost site. In each panel from left to right Kimberley (KRS) (at 10 %), Malapa (MRS) (at 30 %), Gladysvale (GRS) (at 35 %). Percentage C3 axis has been truncated from 0-50 % for clarity. A) All taxa; B) Rodents only; C) Dominant Rodent Taxa (Micaelamys, Mastomys, and Otomys); D) Dominant Generalist Rodent Taxa (Micaelamys and Mastomys only).

62

3.3.4 Fossil Analyses

A major goal of this study was to explore how the relationship between small

mammal enamel stable carbon isotope values and habitat composition could be used in

paleoenvironmental reconstruction. It is clear from the discussion above that such

relationships are complex, but our results suggest that small mammal carbon isotope

compositions can, with appropriate taxonomic control, reveal information about ancient

habitat structure in mosaic environments. Given this, we analyzed the carbon isotope

compositions of micromammals from Sterkfontein (Sts), Swartkrans (Sk), and

Gladysvale (GV) fossil sites in the Sterkfontein Valley. The carbon isotope values of

small mammal teeth from the three fossil sites differ significantly (Welch’s F2,43.633 =

10.461, p = 0.0002). Those from Swartkrans ( = –8.8 ‰) have significantly lower

mean δ13C values than those of both Sterkfontein ( = –6.0 ‰) and Gladysvale fossil

sites ( = –6.1 ‰) (Tukey’s HSD, p < 0.001) (see Figure 3.6). 63

0 -1 -2 -3 -4 -5 -6 -7 -8 C ‰ (PDB) 13 δ -9 -10 -11 -12 -13 -14 -15 SterkfonteinSterkfontein ( StsM4) SwartkransSwartkrans (M1Sk) GladysvaleGladysvale Fossil(GV) (2.6 – 2.2 Ma) (~ 1.8 Ma) (~ 700 Ka)

Figure 3.6: Box-and-whisker plots of d13C by fossil site. 64 These results are interesting in light of published reconstructions based upon both taxonomic, ecomorphological (e.g., Vrba, 1975, 1985; Reed, 1997), and stable isotope analyses

(Lee-Thorp et al., 2007) from hominin sites in the Sterkfontein Valley. Specifically, Swartkrans

M1 has frequently been reconstructed as more open and C4 - dominated than Sterkfontein M4

(e.g., Vrba, 1975, 1985; Reed, 1997; Lee-Thorp et al., 2007). However, our isotopic analyses yielded a relatively low mean d13C value for small mammals at Swartkrans, possibly indicating high C3 vegetation availability. At first glance, these results might lead one to characterize the environment at Swartkrans as relatively more closed and C3 dominated. However, in light of the data presented earlier in this paper, such an interpretation is probably premature. Mean d13C values obtained from modern small mammals in this study did not produce interpretable patterns unless taxonomy was taken into consideration. When fossil sites are broken down by taxon to further investigate patterning within the isotopic dataset (see Table 3.4: Summary statistics for carbon and oxygen isotope data by taxon for each modern roost and fossil site (Mean±SD (N)).) it becomes apparent that the lower mean d13C value at Swartkrans is driven primarily by Otomys sp. In our analyses of modern samples, the most open, grassland dominated roost, Kimberley

(KRS) had unexpectedly low mean d13C values – a result we determined to also be driven by the presence of Otomys sp. As a wetland specialist, the strong C3 signal evidenced by Otomys is likely explained by its lifestyle, since plants growing around permanent water bodies in these

13 environments are often C3 (Happold, 2013). The relatively low d C value at Swartkrans relative to the other sites may be a similar artifact of taxonomic bias. In this instance, the combination of small mammal datasets with other lines of evidence (e.g. analyses of large-bodied mammals) potentially provides a more complete picture of local environment – the latter indicating primarily open habitat, and the former suggesting that permanent water bodies surrounded by C3- 65 dominated vegetation also dotted the landscape. Regardless, the results of our modern analyses suggest that taxonomy is important in interpreting relationships between small mammal isotope ecology and habitat composition.

Finally, we compared the carbon isotope compositions of modern (GRS) and fossil (GV) specimens from Gladysvale Cave. These two assemblages derive from the same cave system but are temporally distinct. There is taxonomic overlap between the two time periods, but it is not complete. For this comparison, an additional 1.5 ‰ was added to the modern Gladysvale Roost

Site (GRS) δ13C values to account for changes in atmospheric δ13C due to the burning of fossil fuels effect and make these data directly comparable (Francey et al., 1999). Figure 3.7: Box- and-whisker plots of small mammal carbon isotope compositions by taxon for Gladysvale fossil

(GV) and Gladysvale modern roost (GRS) sites.illustrates box plots of fossil and modern d13C data by taxon. The Gladysvale fossils (GV) ( = –6.1 ‰) have significantly higher δ13C values than their modern counterparts ( = –8.7 ‰) (Welch’s F1,53.785 = 8.617, p = 0.0024). In addition, the modern data (SD = 4.8 ‰) were significantly more variable than the fossil data (SD = 2.4 ‰)

(F35,30 = 3.8037, p = 0.0003). 66

1 0 -1 -2 -3 -4 -5 -6 -7 -8 -9 -10

C ‰ (PDB) -11 13

δ -12 -13 -14 -15 -16 -17 -18 -19 -20

Otomys Murinae Otomys Mastomys Cryptomys MicaelamysMystromys DendromusGerbillinae Micaelamys Macroscelididae

Fossil Modern

Figure 3.7: Box-and-whisker plots of small mammal carbon isotope compositions by taxon for Gladysvale fossil (GV) and Gladysvale modern roost (GRS) sites. 67 As with any interpretation of a fossil assemblage, both possible taphonomic biases and the assumptions of taxonomic uniformitarianism potentially complicate interpretation of the data.

However, at a general level, the fossils exhibit higher d13C values than their modern counterparts. This indicates a greater contribution of C4 resources to the diets of small mammals,

which likely reflects more C4 grass availability at Gladysvale in the past. Intriguingly, fossil and modern small mammal insectivores from the area exhibit the same pattern (Leichliter et al.,

2016).

CONCLUSIONS

13 Our analyses of modern small mammal d Cenamel composition and habitat suggest that the relationship between community mean d13C and vegetation composition is complex. Inter- and intra-taxon variation is extremely high compared to many large generalist herbivores in similar African ecosystems. Even in seemingly homogeneous savanna environments, small mammals find and consume resources with d13C values ranging from very high to very low. For example, small mammals at the grass-dominated site of Kimberley (95% grass), where higher

13 d C compositions might be expected due to the larger proportion of C4 resources locally available, nonetheless consumed a significant proportion of C3 resources. It is, therefore, apparent that small mammals engage with habitat heterogeneity (mosaicism) at scales that large fauna do not, and that they partition themselves into ecological/dietary niches accordingly. This makes them sensitive proxies for vegetation composition but also makes their carbon isotope compositions more difficult to interpret in a broader sense. While the C3- C4 axis may be a good proxy for open-closed environments in large fauna, it not clear that it will always be as useful for smaller fauna. Taphonomic processes are another crucially important consideration in interpreting carbon isotopic datasets of this type. Here, we used pellet collections from the barn 68 owls which are known to introduce small biases—they consume primarily nocturnal prey species, they can only take prey up to a certain body size, and they have a tendency to hunt in the more open areas of their environments where they can more easily maneuver. Nevertheless, they were near ideal accumulators for our study given our detailed knowledge of their ecology, the fact that they sample the small mammal communities within their hunting ranges with reasonable fidelity (Andrews, 1990; Avenant, 2005; Reed, 2007; Terry, 2010), and because they were the principle accumulators at the southern African hominin sites which we hope to explore with these methods (Avery, 2001). Still, there can be no question that investigation of the relationships between habitat, small mammal diet and distribution, and predator prey-selection

13 require further study. Our study suggests, however, that small mammal d Cenamel values have the potential to reveal a good deal about past habitats, especially when interpreted in light of taxonomic and relative abundance data. 69 4 MICROMAMMAL FOSSIL DESCRIPTIONS

INTRODUCTION

Paleoecological reconstructions utilizing taxonomic analyses rely heavily upon correct identification of taxa in faunal assemblages. Moreover, robust sample sizes are needed both to approximate the diversity of fossil faunal communities and to investigate potential taphonomic biases in a given assemblage. In addition to re-analyzing previously published taxonomic data

(see Chapter 6), > 1,000 new micromammal fossil specimens (minimum number of individuals) were identified for taxonomic and stable isotope analyses. This chapter presents descriptions of micromammalian fossil specimens from four Plio-Pleistocene deposits in the Sterkfontein

Valley. These deposits include Kromdraai B (KB) (~1.8-1.6 Ma), Sterkfontein Member 4 (ST-

M4) (~2.6-2.0 Ma), Swartkrans Member 1 – Hanging Remnant (SK-M1) (2.0-1.8 Ma) and

Gladysvale External Deposits (GVED) (780-570 ka) (Curnoe et al., 2001; Thackeray et al., 2002;

Lacruz et al., 2003; Herries et al., 2009; Pickering et al., 2011; Herries et al., 2013). The largest sample is KB (n=689), followed by SK-M1 (n=158), ST-M4 (n=106), and GVED (n=93).

Overall, 1, 046 specimens belonging to 16 taxa are described.

Fossils derive from prior excavations at each site and specimens are housed at the

Ditsong (formerly Transvaal) National Museum. The deposits from which the fossils derive are all similar in nature – they were accumulated as debris infill in solution cavities which formed in the dolomitic limestone substrate characteristic of the area. More detailed information on each site is provided below (see 4.2).

Only identifiable cranio-dental material is described here, although significant amounts of post-cranial material also exist. The classificatory scheme of Happold (2013) is followed for order, family, subfamily, genus, and occasionally species designations. Specimens were 70 identified using modern and fossil comparative material from the Ditsong National Museum,

Pretoria and the Bloemfontein National Museum, Bloemfontein, South Africa. Fossil specimens were also compared to published images and descriptions of extant (De Graaff, 1981; Skinner and Chimimba, 2005; Happold, 2013) and fossil taxa (De Graaff, 1960; Cooke, 1963, 1990;

Butler and Greenwood, 1976, 1979; Jaeger, 1976, Wesselman, 1984; Pocock, 1987; Denys,

1990; Avery, 1995, 1998, 2001; Reed, 2011). Sample images of select micromammal fossil taxa are presented in a supplementary file.

Anatomical descriptions and dental nomenclature (Figure 4.1) follow Musser (1987) for rodents (Rodentia) and Butler and Greenwood (1976, 1979) for shrews (Soricomorpha) and sengis (Macroscelidea). Maxillae, mandibles, and isolated teeth were identified to the most specific taxonomic level possible, however, only maxillary and mandibular data are presented here. This information was used to calculate the minimum number of individuals (MNI) for each deposit. 71

Figure 4.1: Upper dental nomenclature from Miller (1912), lower dental nomenclature from Misnne (1969). Abbreviations: acc, anterocentral cusp; abc, anterobuccal cusplet; pbc, posterior buccal cusplet; a-bucc, anterobucal cusp; pc, posterior cingulum.

SITE BACKGROUND INFORMATION

4.2.1 Kromdraai B

Kromdraai is located on the southern side of the Blaubank River which cuts through the

Sterkfontein Valley. It is within 2-4 km of Sterkfontein and Swartkrans, while Gladysvale is > 10 km away. Traditionally, Kromdraai has been divided into two distinct deposits Kromdraai B

(KB) and Kromdraai A (KA). KA is commonly referred to as the ‘faunal site’ because it is rich in non-hominin fossil remains. Of the two deposits, only KB has produced hominins, but it is 72 depauperate in other fauna with the notable exception of microfauna. To date, the site has produced 21 Paranthropus robustus specimens including the type specimen for this taxon TM

1517. More recent excavations and re-analyses of the site argue for the presence of Homo as well

(Braga et al. 2017).

KB is a deep, ~ 45 m long fissure infill running east to west. Although initially believed to be limited in north/south extent (1-3 m) recent excavations have greatly expanded the site to

~30 m north/south (Braga et al. 2017). Kromdraai B is further subdivided into two formations by a dolomitic bridge, KB ‘West’ and KB ‘East’. The original opening was probably located in the eastern end of the cave system. The deposit derives from a single talus cone inter-bedded with flowstones. Several cycles of deposition and erosion have resulted in a complex succession of sediments that represent more than a single time period. Currently, KB ‘East’ is comprised of 5 stratigraphic members (M1-5), while KB ‘west’ consists of only 3 members.

Historically, Member 3 has proven the most fossiliferous deposit and is probably the member from which most of the hominin and other faunal material is derived. However,

Bruxelles et al., (2018) assert that it is difficult to unambiguously assign any material removed prior to systematic excavation of the site by Vrba in 1980 to Member 3. The site has gone through 5 primary phases of excavation with the last still ongoing. Most of the material included in this study likely derives from Vrba’s excavations of the central part of the KB East formation in the 1980s. In addition, new material excavated by the Kromdraai Research Project is also described. KB has been dated using faunal analysis and paleomagnetism to ~ 1.8-1.6 Ma and is reconstructed as an open, grassy environment (Vrba, 1975, 1985; Herries et al., 2013; Bruxelles et al., 2017). 73 4.2.2 Sterkfontein Member 4

Sterkfontein Member 4 is also located on the south side of the Bloubank River in the

Sterkfontein Valley and is nearer Swartkrans than is Kromdraai. It is comprised of 6 members, numbered 1-6 from oldest to youngest. The stratigraphy of the site is complex and has been altered by mining operations, which led to the initial discovery of fossils, as well as numerous excavations. Member 4 is a large breccia infill and is considered the original “Type Site”, while

Member 5 is the “Extension Site”. Member 4 has produced over 600 Australopithecus africanus specimens and a rich faunal record (Klein, 2009). No stone tools have been recovered from this member. More recently, U-Pb and U-Th dates obtained for Sterkfontein indicate that Member 4 dates between 2.65 ± 3.0 and 2.01± 0.05 Ma and accumulated over a relatively long period of time (~800 ka). These dates therefore constrain Australopithecus africanus at this site to > 2 Ma

(Herries et al., 2009; Pickering et al., 2011). Analyses of fauna (McKee, 1991; Vrba 1975, 1985) and flora (Bamford, 1999) have been interpreted to indicate a moderately mesic/wooded environment at the time of deposition, though interpretations based on microfauna have differed

(Avery, 2001) (see Chapter 1 for discussion).

4.2.3 Swartkrans Member 1 (Hanging Remnant)

Swartkrans is located opposite Sterkfontein, on the north side of the Bloubank River. The stratigraphy of this site has been extensively studied by Brain (Brain et al., 1988, 1993) and is complex, but well understood. The site consists of 5 members. Member 1 is the oldest and is divided into two distinct deposits, the Lower Bank and Hanging Remnant, which were divided as the result of erosional processes (Klein, 2009). Micromammal specimens described in this chapter all derive from the Swartkrans Member 1 Hanging Remnant. Avery (1998, 2001) does not specify which deposit the SK-M1 specimens she analyzed came from, but states only that 74 they were collected by C.K. Brain in excavations conducted from 1979-1986. Members 1-3 have produced over 300 Paranthropus robustus specimens, the large majority of which derived from

Member 1 (Brain, 1981; Grine, 1993). Hominin specimens referred to the genus Homo, and stone tools have also been recovered from Swartkrans (Grine, 2005; Clark, 1993). Faunal reconstructions suggest that a dry grassy environment prevailed during the accumulation of this deposit (Vrba 1975, 1985).

4.2.4 Gladysvale External Deposit

Gladysvale is located 13 km north-east of Sterkfontein and Swartkrans in the Malapa

Reserve. The site has produced a few potential hominin specimens (Berger, 1993; Berger et al.,

1993), an acheulean handaxe (Hall et al., 2006), and a very large quantity of faunal remains

(Lacruz et al., 2003). Two primary deposits are recognized, the Gladysvale Internal Deposits

(GVID) and the Gladysvale External Deposits (GVED) (Lacruz et al., 2003; Pickering et al.,

2007). Micromammal material described in this chapter derive from the younger GVED deposit.

GVED is 20 x 5 m thick, horizontally stratified, and consists of domes of calcified breccia and pockets of associated decalcified material. The deposits have been divided into 9 sedimentary units (A to I) by excavators but are argued to represent a single deposit (Lacruz et al., 2003). The site was first excavated by Broom in 1936 but systematic excavations did not begin until the

1990s (Berger, 1993; Berger et al., 1993). GVED is estimated to date between 780 – 570 ka

(Lacruz et al., 2003; Pickering et al., 2007). Lacruz et al. (2003) tentatively characterize the environment at the time of deposition as “open and edaphic grassland”.

75 SYSTEMATIC PALEONTOLOGY

Order Macroscelidae Butler, 1956

Family Macroscelididae Bonaparte, 1838

Subfamily Macroscelidinae Corbet and Hanks, 1968

Genus Elephantulus Thomas and Schwann, 1906

cf. Elephantulus sp.

Referred Material – KB: MNI 78 (57 right Mandibles, 78 left mandibles, 2 left maxillae)

ST-M4: MNI 5

GVED: MNI 8

Description – Cranio-dental material consists primarily of mandibular fragments with one or more variably worn teeth in situ. Mandibles are long and slender with a maximum of 10-11 sharp teeth. In most cases, some premolar and/or molar teeth are preserved while the anterior and posterior ends of the mandible are missing. However, when present, incisors and canines appear predominantly small and subequal in size. Premolars and molars possess well-developed cusps and increase in size posteriorly.

Discussion – Material that is referred to the subfamily Macroscelidinae probably belongs to the genus Elephantulus. Pocock (1987) identified four elephant shrew species at Kromdraai B and

Sterkfontein; E. antiquus, E. (Nasilio) brachyrhynchus, and Macroscelides cf. proboscideus, and

Macroscelides proboscideus vagans. (Only the former two were reported at Sterkfontein).

Subsequently, Avery (1998) argued that specimens previously assigned to E. antiquus (and E. nanus) by various researchers (e.g. Broom, 1937; Pocock, 1987) do not differ significantly enough from the modern species E. intufi to warrant unique designations. Re-description of fossil 76 type-specimens have yet to be undertaken to establish E. antiquus as a valid taxon. Avery (2001) identified three species at Sterkfontein and Swartkrans; E. fuscus, E. intufi, and M. proboscideus.

At all sites Elephantulus was much more common than the rest and this appears to be the case for the material described here. However, several of the species belonging to this genus are very similar and are difficult to distinguish with fragmentary material. The specimens described here have been conservatively referred to the family Macroscelidinae but are probably Elephantulus.

Order Rodentia Bowdich, 1812

Family Bathyergidae Waterhouse, 1841

Genus Cryptomys Illinger, 1811

cf. Cryptomys sp.

Referred Material – KB: MNI 72 (23 right mandibles, 15 left mandibles, 6 right maxillae, 3

left maxillae)

ST-M4: MNI 1

SK-M1: MNI 1

GVED: MNI 10

Description – Maxillary and mandibular morphology in the family Bathyergidae is highly distinctive and even heavily worn and/or edentulous specimens are easily identified to family.

Cranial morphology is robust, inter-orbital constriction is notable, infraorbital foramina are small. Five relatively complete maxillae (i.e. both left and right sides intact) are included in this sample, however in almost all cases the anterior, posterior, and zygomatic portions of the crania are missing. Therefore, assessments of certain cranial characteristics relevant to species 77 identification (e.g. dimensions of the infraorbital foramen, shape of the cranial vault, zygomatic- jugal articulation) cannot be made.

In most specimens, between 1 and 4 in situ teeth are present. No more than 4 cheek teeth

(including 1 pre-molar and 3 molars) were observed. Cheek teeth are rooted, strongly hypsodont, have a simple molariform structure, and are typically broader than they are long. Most cheek teeth in this sample have a simple ring shape consisting of a dentine core surrounded by enamel.

Some teeth show indications of a single re-entrant fold in the enamel (particularly the 3 posterior molars) while others possess folds in both the inner and outer (lingual and buccal) enamel surface.

Discussion – It is possible that two taxa, Cryptomys cf. hottentotus and Georychus capensis are represented in this sample. Pocock (1987) identified two species of mole-rat, Cryptomys robertsi and Cryptomys cf. hottentotus natalensis, at Kromdraai B. According to Pocock C. robertsi is distinguishable by relatively its larger molars but narrower incisors while the C. cf. h. natalensis has smaller molars but broader incisors. However, Avery (1998) concluded that C. robertsi did not differ from extant Georychus capensis. Avery (2001) identified two taxa, C. damarensis and

C. hottentotus at Sterkfontein and Swartkrans, but only C. hottentotus at Gladysvale.

However, in our evaluation of previously catalogued material from Kromdraai B, we note that all Bathyergidae specimens are still labeled designated as Cryptomys robertsi, probably reflecting the original identifications made by Broom (1937) and De Graaff (1975). Clearly this material is in need of careful re-assessment so that the proper species attributions can be made.

Morphological characteristics of the teeth can be used to differentiate Cryptomys and

Georychus capensis. In Georychus capensis, cheek teeth typically possess one inner and one outer re-entrant fold in the enamel. In the maxilla, both folds persist until late in life, even in 78 heavily worn teeth. In the mandibular teeth, the inner fold tends to disappear. In contrast, the teeth of Cryptomys are often simple rings, lacking re-entrant folds. However, Cryptomys is a geographically widely distributed, polytypic species and the posterior 3 teeth of Cryptomys frequently possess traces of re-entrant folds in adults. The teeth of young Cryptomys individuals almost always possess inner and outer folds that are eroded with age and wear (pers. obs.).

Therefore, differentiating young Cryptomys from Georychus capensis is not always straightforward when fossil materials are fragmentary.

Some of the teeth in these samples do show traces of a re-entrant fold. However, indications of such folds can occur in Cryptomys and specimens are overall larger than expected for Georychus. For this reason, specimens are referred only to Cryptomys, although closer analyses of the material may reveal species level differences.

79

Order Rodentia Bowdich, 1812

Family Nesomyidae Major, 1897

Subfamily Dendromurinae G.M. Allen, 1939

Genus Dendromus Smith, 1829

cf. Dendromus sp.

Referred Material – KB: MNI 46 (34 right mandibles, 46 left mandibles, 6 right maxillae, 5 left

maxillae)

ST-M4: MNI 5

SK-M1: MNI 2

GVED: MNI 1

Description – The material referred to Dendromus primarily consists of mandibular material with some fragmentary maxillary material and isolated molars. The occlusal surface of the teeth is cuspidate and toothrows are parallel.

Maxillary molars are biserial (i.e. two cusps in each row). M1 has 3 pairs of cusps, M2 has

2 pairs of cusps, while M3 is tiny but usually characterized by 2 cusps which are heavily worn in most specimens when M3 is preserved. M1 is approximately two times the size of M2.

All specimens preserving M1 also possess an accessory lingual cusp (enterostyle) adjacent to the middle laminae. Though rarely preserved, anterior palatal foraminae reach to the

1 middle of the upper first molar when present. M and M1 lack longitudinal enamel crests, differentiating Dendromus from other members of the subfamily Dendromurinae. 80 Discussion - Despite good documentation of Dendromus in many southern African sites of

Pliocene and Pleistocene age, species level classifications are rarely made with any certainty.

Denys (1994) described two new species (D. darti and D. averyi) at Langebaanweg, but this site is considerably older and geographically separate from the deposits under consideration here.

Pocock (1987) simply identified Dendromus sp. at Kromdraai B, though Avery (2001) attributed

Dendromus at Sterkfontein and Swartkrans to D. cf. melanotis. This designation was given primarily on the basis of size and evidence of a reduced posterior cingulum on M1. According to

Happold, “Systematically, Dendromus is one of the most difficult genera of African rodents” and

“the genus as a whole possess a very homogeneous morphology with respect to skull and dentition”. Species identification often relies on a variety of external characteristics including pelage as well as distributional information. Few genetic analyses have been undertaken and the taxonomy of the genus Dendromus is currently under review, therefore it is unsurprising that classifying fossil material is difficult. Specimens identified here are therefore identified only to genus.

Order Rodentia Bowdich, 1812

Family Nesomyidae Major, 1897

Subfamily Dendromurinae G.M. Allen, 1939

Genus Malacothrix Smith, 1834

Malacothrix cf. typica

Referred Material – KB: MNI 4 (1 right mandible, 4 left mandibles 1 left maxilla)

Description – The genus Malacothrix shares many features with other members of the subfamily

Dendromurinae. These include small size, biserial cusp pattern, tooth size decreasing from the 81 first to the third molar and the third molar is quite minute. The single maxillary fragment in this sample (KB121-501) possesses only M1, but this tooth shows strong affinities extant

Malacothrix typica. The cusp pattern is biserial, with three pairs of cusps. A pronounced cusp is present adjacent to the middle lamina on the lingual side of the two primary cusps.

The remaining material conferred to Malacothrix consists of mandibular fragments with

M1 and M2 in situ. These molars are characterized by interconnecting longitudinal crests.

Discussion – Most researchers attribute specimens with this morphology simply to Malacothrix sp. (Lavocat, 1957; De Graaff, 1961; Pocock, 1987). Pocock (1987) suggested that material from

Kromdraai B looked “more primitive” than modern Malacothrix typica but concluded that representative specimens from this site were probably ancestral to modern forms. De Graaff

(1960), tentatively described a new taxon, Malacothrix makapani based upon features of the M1, a designation accepted and expanded upon (including description of M1) by Denys (1990).

However, in both cases sample sizes were extremely small (n = 1). Justification for erecting a new taxon was based primarily on size differences between fossil and extant material rather than morphology. Avery (1998) suggests that M. makapani could be a valid taxon, but nonetheless referred all Malacothrix material from Sterkfontein and Swartkrans to M. typica (Avery, 2001).

Order Rodentia Bowdich, 1812

Family Nesomyidae Major, 1897

Subfamily Dendromurinae G.M. Allen, 1939

Genus Steatomys Peters, 1846

cf. Steatomys sp.

82 Referred Material – KB MNI 8 (4 right mandibles, 8 left mandibles, 1 right maxilla)

ST-M4: MNI 1

GVED: MNI 1

Description – Cranio-dental morphology in Steatomys specimens is again very similar to the rest of the Dendromurinae. Cusps are arranged in a biserial pattern, M1 is the largest tooth and M3 is the smallest. However, M3 is not as reduced in size relative to M1 and M2 as it is in the genera of this subfamily. Cheekteeth also tend to become laminate with wear, unlike in Dendromus and

Malacothrix who’s teeth exhibit a distorted, “whorled” pattern with wear. Only one maxilla was identified in the current sample while the remaining material is mandibular. All of the mandibles are quite worn and exhibit a laminate pattern, distinguishing them from Dendromus and

Malacothrix.

Discussion – Brain (1981) referred material from Sterkfontein to Steatomys cf. pratensis. Pocock

(1987) simply referred material at Kromdraai to Steatomys sp., although in text he states that it probably has affinities to extant S. krebsii. Avery (1998) suggests that mandibular material at

Swartkrans shows greatest affinity to S. cf. pratensis based upon, “relatively broad lower molars possessing a singular conule” and a “2nd lower molar with anterolabial cingulum and relatively large posterior cingulum”. However, Avery concedes that these features are highly variable and that a species level assessment probably cannot be made. The taxonomic status of this genus is still under review and species differentiation in extant taxa relies primarily upon number of nipples and body size rather than dental differences (Happold, 2013). It seems prudent, therefore, to confer material in this sample simply to Steatomys sp.

83 Order Rodentia Bowdich, 1812

Family Nesomyidae Major, 1897

Subfamily Mystromyinae Vorontsov, 1966

Genus Mystromys Smith, 1834

Mystromys albicaudatus

Referred Material – KB: MNI 282 (282 right mandibles, 251 left mandibles, 91 right maxillae,

94 left maxillae, 5 complete maxillae).

ST-M4: MNI 66

SK-M1: MNI 107

GVED: MNI 31

Description – The dentition of Mystromys is highly distinctive compared to almost all other taxa in these samples. The cusps of both maxillary and mandibular teeth are arranged in two rows rather than three as in the subfamily Murinae. There are no lateral cusps. Cusps alternate in slanting folds that result in a characteristic “zig-zag” pattern making even very worn teeth quite unmistakable.

In the upper teeth, M1 is characterized by two inner and two outer folds, which nearly converge at the tooth’s axis. M2 has two outer and one inner fold, and M3 two outer folds and two inner folds. In the lower jaw, M1 has three inner folds and two outer folds. M2 has two inner and two outer folds (rather than only one inner fold as in the upper second molar). The extent of the palate and anterior palatal foramina were observable in some of the 5 partially complete maxillae. The palate reaches the level M3, while the anterior palatal foramina reach M1 and are quite long. 84 Discussion – The genus Mystromys is well documented in the southern African fossil record during the Plio-Pleistocene age (Broom, 1937; De Graaff, 1960; Pocock, 1987; Denys, 1990,

Avery 1995, 1998, 2001, 2010; Sénegas et al., 2005; Winkler et al., 2010). Following description by Broom (1937) most Mystromys material has been referred to Mystromys hausleitneri

(alternatively hauslichtneri). In addition to the presence of “small but quite well developed anterior cusps on M2” (Broom, 1937), size is frequently cited as the primary diagnostic criterion for assignation of material to M. hausleitneri (Broom, 1937; Denys 1990a).

However, Avery (1998) found complete overlap in size between fossil Mystromys from

Swartkrans and extant Mystromys albicaudatus. She argues that Mystromys hausleitneri represents only a chronospecies and refers all specimens from Sterkfontein, Swartkrans, and

Gladysvale to Mystromys albicaudatus. She further notes that much of the apparent variation in dental characteristics observed by earlier researchers is simply due to varying degrees of wear and that individual size is frequently influenced by ecology.

A second, dwarf form species of Mystromys (M. darti) is acknowledged at

Langebaanweg and Makapansgat (Lavocat, 1957; De Graaff, 1960). Pocock (1987) referred specimens at Makapansgat to this taxon and assigned it a new genus (Stenodontomus darti). No specimens belonging to this dwarf form of Mystromys were identified in these samples.

Order Rodentia Bowdich, 1812

Family Nesomyidae Major, 1897

Subfamily Mystromyinae Vorontsov, 1966

Genus Proodontomys Pocock, 1987

Proodontomys cookei 85

Referred Material – KB: MNI 13 (13 right mandibles, 7 left mandibles, 1 complete maxilla)

Description – Molars are strongly hypsodont with weak cusps and a relatively plain occlusal surface. The teeth are similar to Mystromys in their simplified cusp pattern, which lacks embellishment or lateral cusps, however the toothrow is smaller in size. Molar cusps also differ from those of Mystromys in that they are opposite rather than alternate. At first glance, the teeth appear superficially similar to those of a gerbil because they wear in a manner that makes them appear laminate. According to Pocock (1987) the metacone and hypocone of M1-2 unite early in wear resulting in a simple trilophodont pattern in M1 and bilophodont pattern in M2.

Discussion – Davis (1981) recognized Mystromys cookei in an early manuscript on the microfauna of the Sterkfontein Valley but did not formally describe the species. Pocock (1987) erected a new genus and subsequently transferred specimens previously conferred to M. cookei to Proodontomys cookei. Proodontomys cookei is now recognized at several southern African fossil sites but is notably rarer at younger sites such as Gladysvale (Avery, 1995) and absent entirely from the Holocene deposits at the Cave of Hearths.

Order Rodentia Bowdich, 1812

Family Muridae Illiger, 1811

Subfamily Deomyinae Thomas, 1888

Genus Acomys Geoffroy, 1838

cf. Acomys sp.

86 Referred Material – KB: MNI 2 (2 left mandibles)

SK-M1: MNI 1

Description – Referred material includes two left mandibles, both with M1 in situ. This lower molar possesses two well differentiated anterior cusps separated by a deep groove and two weakly differentiated posterior cusps. M1 also has a reduced terminal heel. Both the mandible and the tooth are similar to Mus in size.

Discussion – Acomys has been well documented from fossil sites in both eastern and southern

Africa (Denys, 1990a; Avery, 1998) and modern taxa are widely distributed in arid to semi-arid regions of Africa (Happold, 2013). Two taxa occur in the southern African subregion, Acomys subspinosis, which is restricted to the Cape, and Acomys spinosissimus. Most Pleistocene fossil

Acomys are not identified to species, and appear indistinguishable from A. spinosissimus (De

Graaff, 1981; Pocock, 1987; Avery, 1998). However, Denys (1990a) referred material from

Makapansgat, Sterkfontein, and Kromdraai A to Acomys mabele, and argues that this fossil taxon is distinguishable from extant Acomys spinosissimus species by the shape of the prelobe on M1.

No such pre-lobe was observed in the described material. Although much work has been done in an effort to clarify the taxonomy of Acomys, Happold (2013) states, “regrettably there is still no comprehensive revision of the genus”.

Order Rodentia Bowdich, 1812

Family Muridae Illiger, 1811

Subfamily Gerbillinae Gray, 1825

Genus Gerbilliscus Thomas, 1897

cf. Gerbilliscus sp. 87

Referred Material – KB: MNI 20 (20 right mandibles, 17 left mandibles, 12 right maxillae, 3

left maxillae)

ST-M4: MNI 5

GVED: MNI 4

Description – Referred material includes both maxillary and mandibular material with at least one molar in situ. Upper incisors possess a single groove and lower incisors are ungrooved. In general, molars are rooted, and cheek teeth exhibit a lophate occlusal pattern. Molars consist of elongated oval rings of enamel surrounding a central cementum and lack clearly the defined cusps characteristic of Murinae. All specimens in this sample are moderately to severely worn and therefore appear laminate (though not to the degree observed in Otomys). The first molar is the largest, the second molar is smaller, and the third molar is the smallest.

For maxillary dentition M1 have three laminae, M2 have two laminae, and M3 have two laminae. M3 sometimes possess only a single lamina with a small posteriorly placed additional lamina, which is offset at an angle from the transverse axis of the first lamina. Mandibular dentition is similar, except that M3 consists of a single lamina and is cylindriform.

Discussion – It is possible that two genera of gerbil are represented in the current sample. Both

Desmodillus and Gerbilliscus were reported Pocock (1987) and Avery (1995, 1998, 2001) at

Kromdraai, Sterkfontein, Swartkrans, and Gladysvale. Species belonging to the genus

Gerbilliscus were previously referred to the genus Tatera (Happold, 2013). Modern taxa endemic to southern Africa include Gerbilliscus leucogaster, Gerbilliscus brantsii, and

Gerbilliscus afra (though the latter is restricted to the Cape region). Gerbilliscus is well represented in the Pleistocene fossil record in southern Africa. Brain (1981) referred similar 88 material from Swartkrans to T. robinsoni, but never formally described the species, rendering this taxon invalid. Representatives of this genus have been identified at Makapansgat,

Sterkfontein, Taung, Kromdraai B, Swartkrans, and Cave of Hearths (Lavocat, 1957; De Graaff,

1960a; Pocock, 1987). Pocock (1987) referred material at Kromdraai B to Tatera (cf.

Gerbilliscus) leucogaster without further comment. However, Avery (1998) found considerable overlap in various molar dimensions between gerbil specimens from Swartkrans and samples of

G. leucogaster and G. brantsii, calling into question the reliability of species level identification in fossil samples referable to this genus. Indeed, Avery (1998) states that, “species belonging to this genus are difficult if not impossible to differentiate based on dental characteristics alone”.

Most specimens in this sample are larger than expected for Desmodillus and morphology is most similar to that of Gerbilliscus. Additional material is needed to clarify the diversity of gerbil species present. In any case, Gerbilliscus is more common than Desmodillus in the fossil deposits in in the region and the same is likely true for the material examined here.

Order Rodentia Bowdich, 1812

Family Muridae Illiger, 1811

Subfamily Murinae Illiger, 1811

Genus Aethomys Thomas, 1915

cf. Aethomys sp. and Aethomys (Micaelamys) cf. namaquensis

Referred Material – cf. Aethomys sp.

KB: MNI 6 (1 right mandible, 6 left mandibles, 6 right maxillae, 1 left

maxilla) 89 ST-M4: MNI 1

SK-M1: MNI 4

Aethomys (Micaelamys) cf. namaquensis

KB: MNI 24 (15 right mandibles, 24 left mandibles, 4 right maxillae, 2

left maxillae)

SK-M1: MNI 3

GVED: MNI 10

Description – As is consistent with the dental characteristics of the subfamily Murinae, all specimens referred to this group possess three cusps in the anterior row of first upper molar. The first molar is the largest and teeth decrease in size distally. Molar cusps tend to be moderately heavy and angular compared to other murine taxa.

In the maxillary dentition, the anterior palatine foramina extend to the first root of M1, although this feature is infrequently preserved in the present collection. M1 is not markedly distorted (unlike in Mastomys sp.) but is instead only very moderately distorted. The central cusps of the tooth are noticeably enlarged and t7 is absent. In specimens most likely referable to the genus Aethomys, M3 is nearly subequal to M2. This is in contrast to specimens probably referable to the genus Aethomys (Micaelamys) namaquensis in which M3 is clearly smaller than

M2. This differentiation is obviously not possible when M2 or M3 are missing.

Aethomys (Micaelamys) namaquensis is more easily differentiated in characteristics of the lower dentition. Specifically, members of this taxon possess a prominent anterior median cusp on M1 resulting in three clear cusps in the anterior row. Other Aethomys species lack this feature and instead only possess two cusps in the anterior row. There is, however, often a small tubercle or cingulum on the lingual surface of M1 in Aethomys, which can confuse identification. 90 Specimens in this analysis have been referred to Aethomys (Micaelamys) namaquensis only when an anterocentral cusp was clearly present and positioned medially within the anterior row of cusps rather than on the front enamel surface of the tooth.

Discussion – Two subgenera, Aethomys and Micaelamys, have long been recognized within this genus. In many classificatory schemes, Micaelamys is elevated to the rank of genus (Skinner and

Chimimba, 2005) and in others it is referred to as Aethomys (Micaelamys) namaquensis (De

Graaff, 1981), or simply Aethomys namaquensis (Happold, 2013). The genus Aethomys has been retained for this analysis.

Morphological characters, namely the smaller relative size of M3 to M2 in Aethomys

(Micaelamys) namaquensis are helpful in differentiating this taxon from Aethomys in cases where both teeth are preserved. Similarly, Aethomys (Micaelamys) namaquensis and Aethomys can be differentiated on the basis of the presence of the additional cusp in the anterior row of M1 in Micaelamys. Moreover, Aethomys frequently possess an anterior tubercle or cingulum, which can be prominent enough to confound positive identification, particularly in fossil specimens.

Regardless, classification based on dental anatomy alone is difficult within this group because there is significant variability within and overlap between taxa.

Aethomys was well established in the southern African subregion by the early Pliocene.

Denys (1990) described two species, A. adamanticola, and A. modernis from Langabaanweg

(~5Mya), the latter of which resembles the extant taxon A. chrysophilis quite closely. This genus has also been identified at Makapansgat, Sterkfontein, Swartkrans, Kromdraai, Glaysvale and

Cave of Hearths (De Graaff, 1960; Denys, 1990a; Pocock, 1987; Avery, 1998, 2001). In most cases this material was referred to either A. chrysophilus and A. namaquensis.

91 Order Rodentia Bowdich, 1812

Family Muridae Illiger, 1811

Subfamily Murinae Illiger, 1811

Genus Dasymys Peters, 1975

cf. Dasymys sp.

Referred Material – KB: 1 (1 left mandible)

Description – One complete left mandible with M1-3 in situ. Teeth are heavily cusped and the center row of cusps including t2, t5, and t8 are broader than lingual and buccal cusps. M1 and M2 lack a terminal heel. There is a large supplementary cusp, as is sometimes described, between and in front of the first two cusps of M1.

Discussion – Multiple fossil species including D. broomi, D. brevirostrus, and D. bolti have been recorded in the Sterkfontein Valley deposits. Avery (1995) concluded that two species are represented at Swartkrans, one of which is probably referable to modern D. incomptus and the second of which represents an as yet undescribed fossil species. Pocock (1987) identified

Dasymys sp. at Kromdraai B, Kromdraai A, and Sterkfontein Member 5, but did not give a species. This specimen is also referred to Dasymys sp.

Order Rodentia Bowdich, 1812

Family Muridae Illiger, 1811

Subfamily Murinae Illiger, 1811

Genus Mastomys Thomas, 1915

cf. Mastomys sp. 92

Referred Material – KB: MNI 6 (5 right mandibles, 6 left mandibles, 4 right maxillae, 4 left

maxillae)

ST-M4: MNI 1

SK-M1: MNI 1

GVED: MNI 2

Description – Molars are well-cusped and of medium size for a murine. When preserved, the anterior palatine foramina extend well between the molar row, usually to the level of the medial root of M1 and never past the first row of cusps in M2. The most distinctive tooth is M1 in which t1 is displaced backwards almost in line with t5 of the second row of cusps. This results in a distinctive “L” shape in the antero-lingual aspect of the first molar in occlusal view. The basal portion of t2 projects forward, but not as dramatically as in the genus Mus. t1 is larger than t3 in all specimens. t7 is lacking. M1 usually (but not always) has three roots which are prominent in most specimens. When all teeth are present, M2 is smaller than M1 and t2 and t7 are absent. M3 is smaller than M2 but is still clearly multi-cusped in less worn/younger specimens.

Mandibular material is more difficult to identify for this taxon. M1 has two large roots. As in Aethomys the anterior four cusps of M1 wear into a “clover leaf” pattern. However, all mandibles are smaller than the size range described for Aethomys and larger than Mus.

Discussion – Multimammate mice are common in both fossil and modern ecosystems throughout

Africa. Both Jaeger (1976) and Wesselman (1984) described a small fossil species with primitive

M1 features, Mastomys minor, at Olduvai and Omo. Reed (2011) describes the tooth morphology of this taxon as similar to that of southern African fossils, but smaller in size. Both Pocock

(1987) and Denys (1990) refer specimens this morphology to the genus Praomys sp., which 93 reflects the taxonomic classificatory scheme being used for the genus at the time, but both decline to give a species designation. Avery (1998), found no statistically significant difference between fossil Mastomys specimens from Swartkrans and modern Mastomys from the

Sterkfontein Valley. Although two Mastomys species M. coucha and M. natalensis exist in southern Africa today even these species can only be confidently differentiated with genetic information (Skinner and Chimimba, 2005; Happold, 2013). Avery (2001) identifies specimens at Sterkfontein and Swartkrans only as Mastomys sp. Specimens described here are similarly conferred to Mastomys sp.

Order Rodentia Bowdich, 1812

Family Muridae Illiger, 1811

Subfamily Murinae Illiger, 1811

Genus Mus Linnaeus, 1758

cf. Mus sp.

Referred Material – KB: MNI 23 (23 right mandibles, 12 left mandibles, 1 right maxilla)

ST-M4: MNI 1

Description – Specimens referred to Mus are distinguishable by their very small size and derived

1 1 tooth morphology. All specimens have at least M or M1 in situ. In the maxillae, the cusps of M are “distorted” such that t1 of the first row of cusps is displaced backwards to such a significant degree that it appears in line with t4 and t5 of the second row. Additionally, t2 is large and the anterior face of the cusp bulges such that the anterior root of M1 is obscured. 94

In the mandible M1 possess 6 clear cusps. The 4 anterior cusps are large and 2 posterior cusps are small. The anterolingual cusp is significantly enlarged and tilted so that it angles buccally and posteriorly.

Discussion – Mus is a common and geographically widespread genus. Although the dental morphology of this genus is distinctive, taxonomy within this group remains complicated, particularly for fossil taxa. Many sources recognize an African subgenus (variably, Nannomys or

Leggada) but this has never been formally recognized.

The oldest documented occurrence of Mus is in eastern Africa. In southern Africa, Mus has been identified at Bolt’s Farm, Makapansgat, Sterkfontein, Taung, Swartkrans, and

Kromdraai. Broom listed Mus material from Bolt’s Farm as Leggada (M.) major, but never formally described the taxon. Avery (2001), referred specimens from Sterkfontein and

Swartkrans to M. cf. triton. De Graaff (1960) referred material from Kromdraai to Mus cf. leggada, but Pocock (1985) declined to specify a species and simply referred material to Mus sp.

Given morphological overlap in dental traits and the generally uncertain taxonomic status of this genus, this analysis follows Pocock (1987) in simply assigning material to Mus sp.

Order Rodentia Bowdich, 1812

Family Muridae Illiger, 1811

Subfamily Murinae Illiger, 1811

Genus Rhabdomys Thomas, 1916

cf. Rhabdomys sp.

95 Referred Material – KB: MNI 7 (4 right mandibles, 1 left mandible, 2 right maxillae, 7 left

maxillae).

SK-M1: MNI 1

Description – Rhabdomys possess a very generalized murine cusp pattern and their teeth lack many unique dental features. They are therefore somewhat difficult to differentiate from other murine taxa. Overall, cusp pattern is similar to that of Aethomys and Lemniscomys, but the teeth are smaller. Cusps are prominent and cone-shaped (bunodont), particularly in the center row of cusps.

Maxillary specimens are comparable in size to Mastomys, but do not exhibit the displacement of t1 (i.e. “L-shaped” antero-lingual aspect) which characterizes Mastomys specimens. t9 of M1 and M2 is reduced. M3 is smaller than M2 but again, not to the degree that it is in Mastomys.

In the maxillary dentition, the two anterior cusps of M1 form an obtuse angle and no anterocentral cusp is present. Similar to the wear pattern observed in Mastomys, a “clover leaf” pattern is evident in the anterior four cusps of M1. When present, M2 is broader than M1. M3 is also broad, but not as broad as M2.

Discussion – Rhabdomys has been recorded from multiple early hominin sites including

Makapansgat, Sterkfontein, Swartkrans, Kromdraai, and Gladysvale (Draper, 1895; De Graaff,

1960; Cooke, 1963, 1990; Avery, 1995, 1998, 2001, 2010). Pocock (1987) lists only Rhabdomys sp. at Kromdraai, Sterkfontein, and Makapansgat, but states that the fossil form represented at these sites looks “primitive” and is probably ancestral to modern forms. However, Avery (1998) found no difference between fossil forms at Swartkrans and the extant species R. pumilio and conferred specimens at Sterkfontein and Swartkrans to this taxon. In general, Rhabdomys are 96 extremely widespread and morphologically variable. They are catholic in their habitat preferences and exhibit high dietary flexibility. Rhabdomys is also diurnal and is therefore somewhat less common than nocturnal taxa in assemblages accumulated by owls (Skinner and

Chimimba, 2005; Happold, 2013). The taxon R. pumilio has recently been divided based on genetic analyses (Rambau et al., 2003; Castiglia et al., 2012). However, these species cannot be differentiated on the basis of dental morphology alone. Specimens are therefore only referred to

Rhabdomys sp. in this study.

Order Rodentia Bowdich, 1812

Family Muridae Illiger, 1811

Subfamily Murinae Illiger, 1811

Genus Otomys Cuvier, 1824

cf. Otomys sp.

Referred Material – KB: MNI 72 (62 right mandibles, 72 left mandibles, 9 right maxillae, 6

left maxillae)

ST-M4: MNI 15

SK-M1: MNI 38

GVED: MNI 10

Description – The incisors and cheekteeth of all specimens referred to this genus are highly distinct and easily differentiated from other groups. The molars are hypsodont and lack true cusps resulting in a uniquely derived lamellar pattern. Teeth lie so closely together that they 97 3 almost appear to be a single laminate tooth. M1 and M are notably elongate and possess the greatest number of lamina compared to the other molars.

In the maxillary dentition, the M1 consistently have 3 laminae, M2 have two laminae, and

M3 is somewhat variable with 5-7 laminae. Most specimens in this sample possess six laminae on M3. Laminae are relatively subequal in size for much of the length of the toothrow but decrease in size distally with the final lamina of M3 frequently appearing reduced and semi- circular.

In the mandibular dentition, the M1 have four laminae, M2 have 2 laminae, and M3 have 2 laminae. The buccal side of the first laminae of M1 is angled posteriorly but the orientation of the remaining laminae becomes more transverse distally, with the final laminae of M3 curving slightly. The anterior margin of the first laminae of M1 occasionally has a bi-lobed appearance.

Discussion –There is good evidence that the Otomyinae evolved from murine rodents sometime during the late Miocene. Primitive forms believed to be ancestral to modern lineages have been described including, Eurotomys pelomyoides from Langebaanweg and Euryotomys bolti from

3 Bolt’s Farm. In both of these fossil taxa M and M1 are distinctly elongated yet possess clear cusps rather than the laminae of extant Otomys. The first true Otomys first appeared in southern

Africa at ~ 3.5 to 3 Ma and only subsequently in eastern Africa (i.e. O. petteri ~ 2–1.5 Ma). It is believed that the group originated in southern Africa and dispersed northward (Winkler et al.,

2010).

Members of this group possess highly distinctive tooth morphology which probably evolved as an adaptation to an herbivorous diet. Extant species are differentiated primarily on the basis of number of laminae in M3, number and characteristics of grooves on the incisors, shape of the nasal bones, and the structure of the petrotympanic foramen (De Graaff, 1981; Avery 98 1998; Skinner and Chimimba, 2005; Happold, 2013). The latter two characteristics are, unfortunately rarely preserved and incisors have frequently fallen out. Thus, most species diagnoses depended primarily upon the number of lamina in M3. In general, there has been an increase in number of laminae within members of this genus through time. Pocock (1987) assigned material from Makapansgat, Sterkfontein, and Kromdraai to Otomys cf. gracilis, which was originally described by Broom (1937). Both Broom and Pocock argue that Otomys from the

Plio-Pleistocene are on average smaller in size, possess narrower incisors, and have fewer

(generally five) laminae in M3 than any extant member of this genus.

However, Avery (1998) found little compelling reason to retain this fossil taxon, citing considerable morphological variability within extant species and suggesting that the degree of wear likely influences the number of observable laminae. Indeed, there were instance in this sample when the final lamina and the lamina immediately preceding it were only barely discernable and had merged almost entirely merged with wear. Avery referred specimens from

Sterkfontein and Swartkrans to O. saundersiae and even Pocock (1987) concedes that O. cf. gracilis bears heavy resemblance to O. saundersaie in size and morphology. Most of the specimens included in this analysis possess six, rather than five laminae in M3. They are referred to simply to Otomys sp.

Order Soricomorpha Gregory, 1910

Family Soricidae Fischer, 1815

Subfamily Crocidurinae and Myosoricinae Hutterer, 2005

cf. Crocidurinae and Myosoricinae

99 Referred Material – KB: MNI 22 (16 right mandibles, 22 left mandibles, 1 right maxilla, 14

complete maxillae)

ST-M4: MNI 4

GVED: MNI 3

Description – Soricid teeth are easily differentiated from those of rodents in their lack of ever- growing incisors and diastema. Teeth are sharp, with high pointed cusps. Soricids are differentiated from Chrysochlorids (golden moles) by their dilambdont rather than zalambdont pattern. Mandibles have distinctive doubled condyloid processes, but these processes are not preserved in all specimens. There are six (sometimes seven) teeth in each mandible and number of teeth in the upper jaw varies by genus.

In the maxillary dentition, the first incisor (I1) is prominent and hook-shaped with a notch and secondary cusp anteriorly (which almost resembles a second tooth). This is followed by 2-3 unicuspid teeth and three molars. Members of the genus Myosorex and Suncus possess a tiny extra pre-molar behind I2 while Crocidura lacks this feature. M3 is always the smallest of the molars.

I1 is tilted horizontally and elongate. The buccal enamel of this tooth extends well below

I2, and the root of this tooth is located below M1. Two (occasionally three) unicuspid teeth are located behind I1. I2 “fits into” the lingual groove of I1 and is generally longer than the second unicuspid tooth, P4. The main cusp of P4 (protoconid) is higher than in I2. In Myosorex, a tiny additional unicuspid (P3) is present behind I2. Suncus and Crocidura lack this diminutive tooth.

Unicuspid teeth are followed by three dilambdodont molars. The crown of each of these teeth is broadest posteriorly and the trigonid is triangular in structure. M3 is the shortest and narrowest of the mandibular molars and possess less elevated cusps. 100

Discussion – Multiple soricid taxa have been identified in the Sterkfontein deposits and include species belonging to the genera Myosorex, Suncus, and Crocidura (Broom, 1948; Meester, 1955;

Repenning, 1967; Meser and Meyer, 1972; Butler and Greenwood, 1979; Pocock, 1987; Butler,

1998; Avery 1995, 1998, 2001, 2010). Pocock (1987) reports only Mysorex sp. at Kromdraai and

Sterkfontein. Avery (1998, 2001) identified Myosorex tenuis, Crocidura silacea, and Suncus varilla at Sterkfontein and Swartkrans. At Gladysvale, Avery (1995) recognized Myosorex sp.,

Crocidura sp., and Suncus varilla. All taxa represented in the southern African fossil deposits belong to the sub-families Crocidurinae and Myosoricinae (i.e. white-toothed shrews) and family

Soricidae.

Genera are differentiated primarily on the basis of size, the number of unicuspid teeth in the maxilla or mandible, and the crown pattern of P4 (Repenning, 1967; Butler and Greenwood,

1976; Skinner and Chimimba, 2005; Happold, 2013). However, each genus is quite speciose

(particularly Crocidura) and there is overlap between features, which complicates diagnoses depending on which portions of each specimen are preserved. At this time, specimens have been identified only to the family Soricidae until more detailed analyses can be undertaken.

101 5 AFRICAN SMALL MAMMAL TAXONOMIC HABITAT INDEX

INTRODUCTION

A taxonomic habitat index (THI) is a model which incorporates data on the habitat associations of all fauna within a given community in a systematic way to produce a composite representation of the environment from which the community derives (van Couvering, 1980;

Simpson, 1960; Nesbit-Evans et al., 1981; Fernandez-Jalvo et al., 1998; Reed, 2003, 2007; Nel and Henshilwood, 2016; Nel et al., 2017). This analytical method was initially developed by

Nesbit-Evans et al. (1981) using mammalian fauna from eastern Africa to investigate the paleoecology of Miocene fossil sites in Africa because the authors found the methods available at the time unsatisfactory. For example, Nesbit-Evans et al. (1981) argue that indicator taxa (i.e. taxa that define a trait or characteristic of the environment) which are frequently employed in faunal analyses to identify particular aspects of habitats, are of limited use in highly heterogeneous environments like the savannas of Africa because the best indicator species include taxa highly adapted to habitat extremes (i.e. forest or desert). Habitats intermediate between these extremes (such as African savannas) are difficult to characterize on the basis of indicator taxa alone. Instead, Nesbit-Evans et al. (1981) drew on methods used by modern ecologists and existing models used for fossil analyses (Fleming, 1973; Andrews et al., 1979; van Couvering 1980; Dodd and Stanton, 1990) to develop a tool for paleoecological reconstruction based upon whole fauna rather than indicator species alone. Nesbit-Evans’ model better reflects the flexibility of habitat use in most animals and provides a hierarchical system for weighting different taxonomic levels.

Although originally designed for large fauna, Nesbit-Evans et al.’s (1981) taxonomic habitat index approach is easily adapted for smaller fauna. However, only a few studies have 102 applied THI to small mammal assemblages - primarily in eastern Africa (Fernandez-Jalvo et al.,

1998; Reed, 2003; but see Nel and Henshilwood, 2016; and Nel et al., 2017). To date, THI analyses (sensu Nesbit-Evans et al., 1981) have not been applied to micromammal assemblages from early hominin deposits in southern Africa. Avery (2001) used analogous, but different methods to investigate the small mammal record at Sterkfontein and Swartkrans, which are difficult to compare directly to THI analyses used in studies of eastern African microfauna.

The primary goal of this chapter is to develop and validate a working THI for small mammals which will be used to investigate the paleoecology of several hominin-bearing and faunal deposits in southern Africa. Specifically, this THI will be used, in combination with other analytical methods (i.e. diversity indices, taxonomic ratios, correspondence analyses), to re- assess Avery’s (2001) interpretation of climatic and habitat conditions in the Sterkfontein Valley during the Plio-Pleistocene and to further investigate patterns of habitat change and/or variation over this period.

DEVELOPMENT OF THE TAXONOMIC HABITAT INDEX

To develop a taxonomic habitat index, niche models must first be built for each taxon of interest (usually species) in the community or assemblage. This is done by compiling information drawn from research literature, census data, and museum collections about the ecology, abundance, and habitat associations of each taxon. This information is then used to partition the habitat preference of all taxa into one of several pre-determined habitat categories.

Each ‘niche model’ therefore effectively summarizes the varied habitat preferences of a given animal. Theoretically, niche models constructed in this manner are more accurate than more simplistic approaches because they reflect the fact that most taxa occupy multiple habitats. 103 In their original niche models for eastern African fauna, Nesbit-Evans et al. (1981) use five habitat categories (forest, woodland/bushland, grassland, semi-desert, aquatic-swamp). Taxa which occur in more than one of these habitat types were scored according to the strength of their association with each habitat type. As an example, Nesbit-Evans et al. use the African elephant (Loxodonta africana). In their model, they assign the elephant to each of their five habitat categories as follows; “Forest = .33, Woodland/Bushland = .33, Grassland = .23, Semi- desert = .11; Aquatic-swamp = 0.” According to their model, elephants are weighted more heavily in forest and woodland/bushland categories than in grasslands or semi-desert habitats. In addition, each habitat category is assigned a rank (Forest = 5, Woodland/Bushland = 4,

Grassland = 3, Semi-Desert = 2, Aquatic-swamp = 1), such that a ‘niche index’ value is obtained for each taxon by multiplying the habitat category score for taxon by the ranked value of the category and summing the results. According to Nesbit-Evans et al. niche model, the niche index value for the African Elephant is 2.25 = (.33 x 5) + (.33 x 4) + (.23 x 3) + (.11 x 2) + (0 x 1). The end result is that taxa associated with more arid habitats have lower niche index values (closer to

1) while taxa associated with more mesic/forested habitats have higher niche index values (closer to 5).

For THI analysis, niche models for each taxon in a faunal community are organized in a contingency table with habitat categories across the columns and taxa down the rows (Table A.1 for all small mammal niche models developed for this study). Scores for each taxon are then summed by habitat category (columns) and divided by the total number of taxa in the assemblage to produce predicted habitat proportions based on the entire community. These proportions are then used to generate a stacked histogram, referred to as a ‘habitat spectrum’. This figure depicts the relative proportions of each habitat type and can be compared visually across different sites. 104 Ideally, THI-based habitat reconstructions account for every species in an assemblage.

However, such specificity is rarely possible in fossil assemblages. A major advantage of Nesbit-

Evans et al.’s niche model is that it can be weighted in a hierarchical manner according to taxonomic category (e.g. genus). To do this, the individual scores within each habitat category for all species belonging to the category of interest are added together and divided by the total number of species in the category. This method can be extended to each successive taxonomic level (e.g. sub-family, family) by averaging the scores of taxa contained within the category.

While not ideal, this level of analysis is nonetheless useful for fossil assemblages when taxonomic identification is difficult (due to poor preservation or taxonomic resolution) or when the organism of interest belongs to an extinct taxonomic group/lineage for which no specific model exists.

For the present study, genus, and in a limited number of instances sub-family (i.e.

Gerbillinae indet., see Table A.1 and Table A.2 for taxonomic information and abbreviations), are used as the primary taxonomic level of analysis. This was necessary for several reasons. For one, positive identification of small mammal species can be problematic, especially within certain groups (e.g. shrews and many rodent genera). This issue is further compounded in the fossil record (Avery, 1998; Winkler et al., 2010). Moreover, the data presented in this chapter and Chapter 5 draw upon a variety of studies in which taxa have been identified with varying degrees of specificity or under different taxonomic schemes. While the use of genus rather than species means that some resolution is lost (because niche models for genera which contain many species with widely varying habitat associations are averaged), comparison at the level of genus allows for maximum taxonomic overlap across datasets. 105 To generate a THI for African small mammals, niche models were developed based on documented habitat preferences for each extant species (Skinners and Chimimba, 2005;

Happold, 2013). Unfortunately, this aspect of the THI model is the most subjective because there is no clear method for assigning habitat category scores. Reed (2003) suggests that more accurate niche models can be constructed using detailed abundance and census data collected in well quantified habitats. However, this type of data is difficult to obtain for African small mammal fauna. Regardless, THI is, at the very least, explicit in how habitat scores are apportioned, adding transparency to the ways in which habitat reconstructions are quantified, and leaving room for refinement. For the present study, niche models were verified by comparison to previously published models for the same taxa (when available) and by evaluating calculated niche index values against published generalized ecological information for each genus.

For small mammmals, a modified version Nesbit-Evans et al.’s (1981) five habitat category scheme was used. This modified habitat scheme was developed by Fernandez-Jalvo et al. (1998) for application to the small mammal assemblages at the eastern African hominin site

Olduvai. Five habitat categories are used, but the “woodland/bushland” category has been split and the “aquatic-swamp” category has been eliminated (Fernandez-Jalvo et al., 1998; Reed,

2003, 2007). Some studies have further modified this model to capture certain elements of vegetation type and substrate (see Nel and Henshilwood, 2016 and Nel et al., 2017). There is significant room for discussion and improvement regarding how habitat categories in THI analyses are defined and which categories are best suited to the community of interest. In both

Nesbit-Evans et al.’s (1981) model for large mammals and Fernandez-Jalvo et al.’s (1998) model for small mammals, for example, habitat categories are defined primarily by vegetation type, but occasionally by amount of precipitation. This inconsistency is clearly problematic and requires 106 revision. However, for the sake of continuity with previous studies of eastern African microfauna, the five habitat categories established by Fernandez-Jalvo et al. (1998) are retained here.

The orders Afrosoricida, Macroscelididea, Rodentia, and Soricomorpha are all included in the THI. Chiropterans are excluded from the model because they are volant and arguably less ecologically informative. Niche models have been generated for taxa which are not considered

“micromammals” (e.g., Pedetes), but which belong to one of the included orders. However, these taxa have not been used in generating habitat reconstructions and are included only for completeness. A benefit of constructing the model in this way is that taxonomic groups can be included or excluded from analyses to evaluate their impact upon the resulting habitat spectra.

This flexibility is useful for making comparisons across datasets. Fernandez-Jalvo et al. (1981) and Reed (2003, 2007), for example, limited their analyses to the order Rodentia, while Avery

(2001) also considered Afrosoricida, Macroscelidea, and Soricomorpha. The THI developed in this chapter allows for easy comparison across these studies. Moreover, all previous microfaunal studies which employ THIs have tailored their niche models to the regions and microhabitats in which the fossil assemblages are located. Fernandez-Jalvo et al. (1998) and Reed (2003, 2007), for example, only incorporate taxa from eastern Africa. However, an explicit goal of the model developed for this study is to facilitate direct comparison across regions. Therefore, all well recognized African species within a given genus were incorporated in order to generate a “pan-

African” model. New niche models are indicated with an asterisk in Table A.1). Niche models for genera occurring in both eastern and southern Africa were produced independently and then compared to those developed by Reed (2003, 2007) for small mammal communities in the

Serengeti. Discrepancies between models are summarized in Table 5.1. 107 Table 5.1: Niche model differences between Reed (2003) and this study Genus Abbreviation Significantly Different DEND Scored significantly higher for grassland STEA Scored "0" for forest GERBC Scored higher for woodland and semi-arid Scored higher for woodland and bushland, lower for semi- LEMN arid MUS Scored higher for woodland and grassland, lower for forest Similar ACOM Scored slightly lower for grassland AETH Scored slightly lower for forest GRAM Scored slightly lower for semi-arid ZELO Scored slightly higher for woodland and bushland OTOM Scored slightly higher for grassland

EVALUATING THE TAXONOMIC HABITAT INDEX

5.3.1 Biozones

In order to evaluate the efficacy of the small mammal THI, presence/absence data (from

Denys, 1999) were used to generate expected habitat proportions for the major biozones of

Africa. Biozones include the Sahara, Sahelian, Sudanian, Guinean, Forest, Montane Forest,

Eastern Forest Savanna (Lake Victoria Basin on Figure 5.1B), Somali-Masai, Zambezian,

Highveld, Namib, Kalahari/South-West Arid, and Cape (SAH, SEL, SUD, GUI, FOR, MON,

EFS, MAS, ZBZ, HIG, NAM, KAL, and CAPE). Figure 5.1A is included to indicate the generalized distribution of habitat types (i.e. forest, desert, etc.) across Africa while Figure 5.1B indicates the boundaries of each biozone. 108

A B

Figure 5.1: Distribution of major biome types (A) and biozones (B) in Africa.

109 Two habitat spectra were generated using the THI for all taxa. The first allows for a coarse-grained evaluation of predicted vegetation using woody vs. non-woody categories (Figure

5.2). These categories are calculated simply by combining the forest, woodland, and bushland categories (woody) and grassland and semi-arid categories (non-woody) respectively from the original five category model. Combining the categories in this way produces a simple, dichotomous approximation of the vegetation predicted and is useful for broadly characterizing environments. Figure 5.3 shows the predicted proportions of the original five habitat categories.

Spectra are organized in order of increasing non-woody or semi-arid categories.

Figure 5.2: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for all Biozones based on the THI for all taxa, using presence/absence data. Biozones are organized in ascending order from left to right by percentage non-woody vegetation. 110

100%

90%

80%

70%

60%

Semi-Arid 50% Grassland Bushland Woodland 40% Forest

30%

20%

10%

0% MON FOR EFS ZBZ HIG GUI CAPE MAS SUD KAL NAM SEL SAH Biozone

Figure 5.3: Habitat spectra for proportions of five habitat categories for all Biozones based on the THI for all taxa, using presence/absence data. Biozones are organized in ascending order from left to right by percentage semi-arid habitat predicted.

In general, habitat spectra reflect biozone habitat compositions well. The Forest, Montane

Forest and Eastern Forest Savanna, and Zambezian biozones consistently appear on the left side of both figures and high proportions of forest/woodland are indicated in each. The semi-arid/arid

Sahara, Sahel, Namib and Kalahari appear to the right of the figure, while mixed savannas and bushland occur in the middle. Some minor discrepancies can be observed, for example, in Figure

5.3 the “forest” proportion of the Sahara biozone seems unexpectedly high. However, this 111 discrepancy is readily explained by the inclusion of taxa from the coastal forest mosaic of the

Mediterranean region in Denys’ (1999) original dataset.

100%

90%

80%

70%

60%

Semi-Arid 50% Grassland Bushland Woodland 40% Forest

30%

20%

10%

0% FOR MON EFS ZBZ HIG GUI MAS CAPE SUD KAL NAM SEL SAH Biozone

Figure 5.4: Habitat spectra for proportions of five habitat categories for all Biozones based on the THI for rodents only, using presence/absence data. Biozones are organized in ascending order from left to right by percentage semi-arid habitat predicted.

When the reconstruction is limited to rodents only (Figure 5.4), the overall effect is an increase in the proportions of some habitat categories and a decrease in others. However, no one habitat category appears to be altered in a consistent direction, suggesting that restricting analyses to rodents does not introduce specific habitat biases into the model. 112 5.3.2 Roost Sites

It is not altogether surprising that the models work well on a continental scale, after all, these areas are defined explicitly by the distinctness of their biotic communities. It is more important to determine whether or not they also work at regional and sub-regional scales. To test the model’s efficacy in the southern African sub-region, habitat spectra were generated for 25 barn owl roosting sites sampled by Vernon (1972) and located throughout southern Africa (see

Figure 5.5). Roost samples have the added benefit of providing taphonomic control for fossil studies because the small mammal fossil assemblages from southern Africa analyzed in Chapter

6 are considered to have been accumulated by barn owls (for detailed discussion of barn owl taphonomy, see Chapter 6).

Figure 5.5: Map of southern Africa showing locations of barn owl roosts included in this study. Map Data: Google, DigitalGlobe. 113

The roost sites analyzed here are distributed across each of the major biomes of southern

Africa (see Figure 5.6). Once again, habitat spectra for woody/non-woody and all five habitat categories appear to reflect the relative proportions of each habitat type expected for each roost fairly well (Figure 5.8 and Figure 5.9).

Figure 5.6: Map of major biomes of South Africa. Map from Mucina and Rutherford (2011).

Roosts 2,4 and 5, all located in the more mesic, eastern part of southern Africa (see

Figure 5.5 and Figure 5.6) and Roosts 26 and 22, located in the highveld grassland, occur on the left side of both figures. Roosts 13, 14, 15, and 19 are all located in the more arid, western part of the region and all consistently appear on the right side of both figures (see Figure 5.5 and Figure

5.6). With the exception of Roosts 26 and 22, most of the roosts located in the highveld (which 114 encompasses the Sterkfontein Valley) fall in the middle of the spectra (see Figure 5.7 for satellite images of representative roost sites).

A B C D

Figure 5.7: Satellite images of; (A) Roost 3, (B) Roost 5, (C) Roost 27, and (D) Roost 19. Google, DigitalGlobe.

115

100%

90%

80%

70%

60%

50% Non-Woody Woody 40%

30%

20%

10%

0% RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST 26 22 2 5 4 21 1 23 20 11 25 12 24 3 27 28 6 10 7 9 13 15 8 19 14 Roost

Figure 5.8: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for all Roost sites based on the THI for all taxa, using presence/absence data. Roosts are organized in ascending order from left to right by percentage non-woody vegetation.

116

100%

90%

80%

70%

60%

Semi-Arid 50% Grassland Bushland 40% Woodland Forest

30%

20%

10%

0% RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST 3 5 2 1 4 26 24 11 23 22 21 27 28 25 20 9 6 7 8 12 15 13 10 14 19 Roost

Figure 5.9: Habitat spectra for proportions of five habitat categories for all roosts based on the THI for all taxa, using presence/absence data. Roosts are organized in ascending order from left to right by percentage semi-arid habitat predicted.

Although the THI produces reasonable habitat predictions using only simple presence/absence data, taxa are almost never equally represented in modern ecosystems. More often, certain species are more common in assemblages, usually in a manner related to the habitat characteristics of the specific environment from which they derive. A more nuanced reconstruction can be generated by accounting for the relative abundance of each taxon. This is done by multiplying the score of each taxon in each habitat category by the abundance of that taxon in the assemblage before summing the scores for each habitat category. 117 The results of weighting the THI for relative abundance are illustrated in Figure 5.10. The general order of the roosts is preserved (roosts 1-5 still cluster at the left side of the figure and roosts 12-14 appear on the right), however, the relative proportions of each habitat category differ. This is particularly the case at more open roost sites. At these roosts, arid-adapted taxa including gerbils dominate the assemblages (see Table A.6) and this is reflected in the increased proportions of semi-arid habitat indicated in the spectra. Limiting analyses to rodents only has a similar effect.

100%

90%

80%

70%

60%

Semi-Arid 50% Grassland Bushland 40% Woodland Forest

30%

20%

10%

0% RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST RST 2 4 3 28 1 5 21 23 24 22 20 27 26 11 25 8 9 6 19 10 15 7 13 14 12 Roost

Figure 5.10: Habitat spectra for proportions of five habitat categories for all roosts based on the THI for all taxa, weighted for relative abundance. Roosts are organized in ascending order from left to right by percentage semi-arid habitat predicted.

118 In all cases, habitat reconstructions which are weighted for relative abundance better reflect the actual vegetation at each roost site. For example, in Figure 5.9 (all taxa, unweighted), approximately 25% forest/woodland is predicted for Roost 6. However, when analyses are weighted for relative abundance, these habitat elements disappear entirely (see Figure 5.10).

These proportions more accurately reflect the actual conditions at Roost 6 (see Figure 5.11).

Figure 5.11: Satellite images of Roost 6. Google, DigitalGlobe

In conclusion, data derived from small mammal community assemblages can be used in combination with the THI developed in this chapter to reconstruct habitat composition.

Presence/absence data can be used to generate predicted habitat proportions but are more likely to indicate the presence of habitat categories which do not actually exist. Reconstructions based upon relative abundance data more accurately reflected habitat at each roost. It does not appear to matter if relative abundance data are limited to rodents only. Reconstructions using all taxa

(weighted for relative abundance) and rodents only (weighted for relative abundance) produce similar results. In all cases, the dominant habitat categories in each biozone and at each roost site were predicted. This model can therefore be applied to the fossil record in southern Africa with 119 reasonable confidence that the dominant habitat elements will be represented, particularly across broad periods of time. 120 6 PALEOECOLOGICAL ANALYSES

INTRODUCTION

The micromammal fossil record from of southern Africa is, in many ways, an ideal dataset with which to test hypotheses about environmental change during early hominin evolution.

Unlike many eastern African hominin localities, the small mammal fossil record in southern

Africa is quite robust and specimens are usually well preserved. In eastern Africa, micomammal assemblages tend to consist of fewer specimens which have often been subject to fluvial transport and sorting. Moreover, the sedimentological context of the fossils requires labor intensive methods such as sieving for recovery of small fossil remains. In southern Africa, small mammal remains usually accumulate and fossilize directly in the cave systems from which they are recovered and acid preparation of fossil breccias produce large amounts of small mammal material. The assemblages considered in this chapter are comprised of hundreds and more often thousands of individuals. Most of the Plio-Pleistocene hominin-bearing deposits in southern

Africa are located within < 20 km of one another and range from ~3 to <1 Ma in age. There is significant taxonomic overlap between sites and between modern and fossil faunas. Finally, there is good evidence that nearly all of the assemblages were accumulated by barn owls and can be considered isotaphonomic (Davis, 1959; De Graaff, 1960; Brain, 1981; Pocock, 1985, 1987;

Andrews, 1990; Avery, 1998, 2001, 2010; Fernandez-Jalvo and Avery, 2016; Sénegas et al.,

2005). It is therefore likely that differences in taxonomic composition, particularly the relative abundance of taxa in each deposit, reflect real ecological differences and not taphonomic, geographic, or taxonomic biases.

This chapter re-assesses Avery’s interpretation of the paleoenvironment in the Sterkfontein

Valley during the Plio-Pleistocene (see Chapter 1 for details). In addition to Sterkfontein and 121 Swartkrans eight additional Plio-Pleistocene deposits have been included in these analyses (see

Table 6.1 for site information and abbreviations). Six of these deposits; Bolt’s Farm, Kromdraai

B, Kromdraai A, Drimolen, Gondolin, and Gladysvale, are located in the Sterkfontein Valley.

Makapansgat, Taung, Wonderwerk, and Langebanweg are located elsewhere in southern Africa

(see Figure 6.1). Makapansgat, Sterkfontein Member 4, and Taung are associated with

Australopithecus africanus, while Sterkfontein Member 5, Swartkrans, Kromdraai B, Drimolen, and Gondolin are associated with Paranthropus robustus. Bolt’s Farm, Kromdraai A,

Wonderwerk, and Gladysvale have all produced rich Plio-Pleistocene faunal assemblages and are included for comparison with the hominin-bearing deposits. 122 Table 6.1: Information about the fossil sites included in this study

Site Abbreviations Site Age Associated Hominin Material References Southern Africa LBW Langebaanweg 5.2 Ma Fauna Only Matthews et al., 2007; Roberts et al., 2011 BF Bolt’s Farm 4.4 - 4 Ma Fauna Only Matthews et al., 2007 TAUNG Taung 3 - 2.6 Ma Australopithecus africanus McKee, 1993; Herries et al., 2009 MAK Makapansgat 3.0 - 2.6 Ma Australopithecus africanus Pocock, 1987 Herries et al., 2013 ST-M4 Sterkfontein Member 2.65 – 2.0 Ma Australopithecus africanus Avery, 2001; Herries et al., 2013; Leichliter, this study 4 ST-M5 Sterkfontein Member 2.0 - 1.5 Ma Paranthropus robustus, early Pocock, 1987; Curnoe, 1999; Avery 2001; Pickering and 5 Homo, Stone artifacts Kramers, 2010; Herries and Shaw, 2011 SK-M1 Swartkrans Member 2.0-1.8 Ma Paranthropus robustus, early Curnoe et al., 2001; Avery 2001; Pickering et al., 2011; 1 Homo, Stone artifacts Herries et al., 2013; Leichliter, this study SK-M2 Swartkrans Member 1.7 - 1.1 Ma Paranthropus robustus Avery, 2001; Batler et al., 2008; Herries et al., 2009 2 SK-M3 Swartkrans Member 1.3 - 0.6 Ma Paranthropus robustus Avery, 2001; Herries et al., 2009 3 KB Kromdraai B 1.8 - 1.6 Ma Paranthropus robustus, early Homo Pocock, 1987; Thackeray et al., 2002; Herries et al., 2009; (?) Leichliter, this study KA Kromdraai A 1.8 - 1.5 Ma Fauna Only Pocock, 1987; Herries et al., 2009 DR Drimolen 2.0 - 1.4 Ma Paranthropus robustus Keyser et al., 2000; Sénegas et al., 2005; Herries et al. 2013 GON Gondolin 1.95 - 1.78 Ma Paranthropus robustus Sénegas et al., 2005; Herries et al., 2006; Adams et al., 2007

WW-11 Wonderwerk 1.1 Ma Fauna Only Matmon et al., 2012; Fernandez-Jalvo and Avery, 2015 WW-12 Wonderwerk 1.96-1.78 Ma Fauna Only Matmon et al., 2012; Fernandez-Jalvo and Avery, 2015 GVED Gladysvale External 0.780 – 0.570 Predominantly Fauna Lacruz et al., 2003; Leichliter, this study Deposits Ma Eastern Africa ARA Aramis 4.4 Ma Ardipithecus ramidus Louchart et al., 2009 HADAR Hadar – Sidi Hakoma 3.42 - 3.24 Ma Australopithecus afarensis Reed and Geraads, 2012 OMO Omo Shungura 3.0 Ma Australopithecus afarensis Wesselman, 1984 Member B FLKNN1-3 Olduvai Middle Bed I 1.86-1.76 Ma Paranthropus boisei Reed, 2005 FLKN1-6 Olduvai Upper Bed 1 1.76-1.75 Ma Fauna Only Reed, 2005 123

Figure 6.1: Locations of the early hominin sites included in this study. Maps from Gibbons et al., (2002) and Stratford et al., (2014).

124 The primary goals of this chapter include 1) incorporating additional micromammalian data from new and existing Plio-Pleistocene sites into paleoecological analyses, 2) evaluating the taxonomic composition, diversity, and taxonomic habitat index at each site, 3) re-evaluating

Avery’s (2001) findings at Sterkfontein and Swartkrans, and 4) investigating whether differences in habitat association in Australopithecus africanus versus Paranthropus robusts and early Homo are indicated in the micromammal record.

Taxonomic information including minimum number of individuals, relative abundance, and presence/absence data are used to evaluate species richness and diversity, calculate similarity indices and taxonomic ratios, generate correspondence analyses, and compare habitat composition using the taxonomic habitat index developed in Chapter 5. At the end of the chapter, habitat spectra for both eastern and southern African microfaunal assemblages from hominin- bearing sites are presented.

6.1.1 Assumptions

6.1.1.1 Taxonomic Uniformitarianism

The concept of taxonomic uniformitarianism necessarily undergirds the analyses presented in this study. The principle of uniformitarianism holds that the ecology of a fossil taxon can be assumed to be similar to that of its closest living relatives. Given this assumption, the faunal analyses presented in this chapter necessarily draw heavily upon modern ecological datasets. Fortunately, there is a large degree of taxonomic overlap between Plio-Pleistocene small mammal fauna and modern taxa, strengthening paleoenvironmental reconstructions.

Several extinct taxa occur in the southern African micromammalian fossil record. Many of these are considered chronospecies or are so closely related to extant species that they are variably assigned by different researchers to either fossil or extant taxa (i.e. Mystromys hausleitneri and 125 Mystromys albicaudatus) (Pocock, 1987; Avery, 1995). Regardless, fossil taxa rarely make up a significant proportion of the small mammal assemblages included in this analysis and the difficulties of interpreting fossil species are mostly avoided by analyzing datasets at the taxonomic level of genus.

6.1.1.2 Taphonomy

The small mammal accumulations in the Cradle of Humankind fossil sites are largely the result of avian predators and small carnivorans, primarily owls (Davis, 1959; De Graaff, 1960;

Brain, 1981; Andrews and Nesbit Evans, 1983; Wesselman, 1984; Pocock, 1985, 1987;

Andrews, 1990; Avery, 1998, 2001, 2010; Reed, 2003, 2005). Caves are commonly favored roosting sites for owls (Davis, 1959; Vernon, 1972; Brain, 1981; Reed, 2003, 2005; Avery, 2001,

2010), which regurgitate the undigested remains of their prey in a dense pellet of fur and bones.

Large quantities of small mammal remains frequently amass below roosts and fossilize under the proper conditions (Glue, 1971; Bunn et al., 1982; Taylor, 1994; Avery, 2001). The excellent preservation (including impressions of fossilized owl pellets), completeness of skeletal elements, sheer density of microfaunal specimens in the cave sites, suggest that vast majority of small mammal remains in the fossil assemblages of southern Africa have been accumulated by roosting barn owls (Tyto alba) (Davis, 1959; De Graaff, 1960; Brain, 1981; Pocock, 1985, 1987;

Andrews, 1990; Avery, 1998, 2001, 2010).

Barn owl dietary ecology has been extensively studied (Coetzee, 1963; Colvin 1985;

Andrews, 1990; Taylor, 1994, Yom-Tov and Wool, 1997; Bunn, 2010). Although some research suggests they are slightly biased towards hunting in open environments, selecting prey of certain body masses, and of taking primarily terrestrial prey (Talyor, 1994; Avery, 2001; Matthews,

2000, 2005; Reed, 2003, 2005, 2007; Avenant, 2005, 2007; Terry, 2010), it has also been well 126 demonstrated that barn owls are highly opportunistic and that the composition of their diets reflect the community structure of the small mammal populations upon which they prey (Avery,

1998; Avenant, 2005; Avenant and Cavallini, 2007; Terry, 2010). Moreover, several studies suggest that the relative abundance and diversity of taxa in owl accumulated assemblages correlate well with characteristics of habitats local to roost sites and vary in predictable ways across habitats (Reed, 2003, 2005, 2007; Terry, 2010). Thus, for instance the diet of an owl hunting in a generally more closed environment will contain different proportions and different types of small mammal taxa than an owl hunting in a more open environment (Bunn et al., 1982;

Reed, 2003, 2007; Terry, 2010).

MATERIALS AND METHODS

Faunal data included in these analyses derive primarily from published information on micromammalian assemblages at each site (Wessleman, 1984; Pocock, 1987; Avery, 1995;

Avery, 2001; Reed, 2007; Sénegas et al., 2005; Matthews et al., 2008; Louchart et al., 2009;

Reed and Geraads, 2012), but also include 1,045 newly identified specimens from ST-M4, SK-

M1 (Hanging Remnant), KB, and GV (see Chapter 3 for detailed descriptions). For published data, different units/deposits considered to be the same age were pooled (e.g. Gondolin A and

Gondolin 6, see Sénegas et al., 2005).

The addition of this new material increases the available sample size by the following proportions; ST-M4 (n=105, 205%), SK-M1 (n= 158, 2.5%), KB (n = 689, 53%), GV (n=93,

27%). The sample size at ST-M4, which is also the smallest of the assemblages in this study, has increased most significantly (from n = 51 to 157) with the addition of these new specimens.

While, the primary focus of this chapter is on southern African fossil assemblages, data from eastern African sites are also included. Only sites/deposits with a sample size > 100 were 127 included. Therefore, all deposits from these eastern African sites are not represented because sample sizes were considered insufficient for comparison to the southern African material.

Taxa were identified to genus when possible, using comparative collections of modern and fossil taxa housed at the Ditsong National Museum, Pretoria, South Africa. Minimum number of individuals (MNI) is calculated based on cranio-dental remains. Relative abundance was calculated by dividing the number of individuals in a given taxon by the total number of individuals in each assemblage. MNI, relative abundance, and presence/absence data for all fossil sites can be found in Table A.3 and Table A.9 in the appendix.

6.2.1 Species Richness and Diversity

Biodiversity indices are cornerstones of ecological analyses. A wide variety of methods to estimate diversity exist. Interpretation of these indices is not always straightforward but in general, such indices are used to identify major events or trends in diversity over time and to indicate environmental conditions (Hammer and Harper, 2006). In small mammals, increases in the number of taxa present in an ecosystem (richness) and the number of individuals per species

(diversity) have been associated with increased precipitation, habitat productivity, habitat structure and complexity, and habitat integrity (Nel, 1975; Abramsky, 1978, 1988; Rowe-Rowe and Meester, 1982; Kotler, 1984; Abramsky and Rosenzweig, 1984; Andrews, 1990;

Rosenzweig, 1992; Waide et al., 1999; Andrews and O’Brien, 2000; Mittlebach et al., 2001;

Reed et al., 2006; Avenant and Cavallini, 2007).

One of the simplest ways to assess the diversity of an assemblage is to quantify the number of species present in a sample. Species richness in small mammal communities has been correlated with precipitation and primary productivity (Abramsky and Rosenzweig, 1984; Waide 128 et al., 1999; Mittlebach et al., 2001; Reed et al., 2006) This is referred to as species richness (S).

In this study, genus has been used to quantify richness. In general, richness increases as a function of increasing sample size, an effect which must be taken into consideration when analyzing fossil samples. To investigate differences in species richness between fossil assemblages, rarefaction curves were generated for each fossil assemblage. Rarefaction curves depict the rate of increase in number of species as sample size increases (i.e. number of species

S(n) as a function of n) (Hammer and Harper, 2006). In general, rarefaction curves rise steeply as sample size increases because many novel species are added, but then begin to level off as sample size continues to increase and fewer new species are encountered. Curves which do not level off indicate that true richness has not been sampled. This usually occurs in small samples.

To account for discrepancies in sample size in the fossil record, all of the larger samples are rarified and standardized to the smallest sample. Southern African assemblages were standardized to the smallest sample size (ST-M4) and standardized, rarified samples were compared statistically using a t-test to evaluate differences in richness between deposits.

The Shannon-Weiner Index (H’ = Spi ln pi) is a frequently used diversity index. The index incorporates both the number of taxa present and the relative abundance of each taxon.

Thus, it takes into account both the richness and the evenness of an assemblage. Low values indicate low diversity while higher values indicate higher diversity. However, because both richness and evenness are used to calculate H’, low evenness resulting from the influence of

2 dominant taxa will cause H’ to decrease. The Simpson Index of Dominance (l = S(p i)), is useful in determining the degree to which an assemblage is dominated by a single taxon. Values close to 1 suggest that one taxon is more common than all others in the assemblage, while lower values 129 indicate that taxa are equally common (Hammer and Harper, 2006). Both indices, as well as several others, were calculated for the fossil assemblages.

6.2.2 Similarity Indices

The taxonomic compositions of fossil sites were compared using two similarity indices to evaluate the assemblage at both a high level (presence-absence) and low level (relative abundance). For high-level comparisons, the Jaccard Index was used. The Jaccard Index is simply the number of shared taxa (M) divided by the total number of taxa (JI = M/(M+N) where

N is the total number of all remaining species) (Hammer and Harper, 2006). For low-level (i.e. relative abundance) comparisons, the Bray-Curtis Index was used. The Bray-Curtis index is defined as BC = S(ui – vi)/ S(ui – vi) where ui and vi represent the sample compositions as two vectors in s-dimensional space, and where s is the number of taxa. The Bray-Curtis index effectively reduces to the complement of the Jaccard Index but uses relative abundance instead of presence absence data (Hammer and Harper, 2006). Both indices were used to generate dendrograms to evaluate the degree of similarity between deposits.

6.2.3 Taxonomic Ratios

Many faunal analyses use indicator taxa, that is, taxa whose habitat tolerances/preference indicate that specific habitat conditions or characteristics to inform paleoenvironmental reconstructions. Examples of indicator taxa in African rodent assemblages, for example, include the Shaggy Rats (Dasymys sp.) which are strongly associated with wetlands and the Acacia Tree

Rats (Thallomys sp.) which, as their name implies, live exclusively Acacia trees and feed on its stems, leaves, and pods. The presence of either taxon in an assemblage are considered robust indicators of wetland and Acacia thicket environments, respectively. 130 However, indicator taxa represent a qualitative rather than quantitative paleoecological tool. Compound indices, such as taxonomic ratios, have greater predictive power. For rodents, the ratio Gerbillinae to Murinae correlates with aridity (Jaeger, 1979; Dauphin et al., 1994;

Fernandez-Jalvo, 1998). In general, gerbils exhibit a number of adaptations to aridity (i.e. derived middle-ear anatomy, concentration of urea, water independence, saltation) and occur in higher frequencies than murines in xeric environments (Skinner and Chimimba, 2005; Happold,

2013). Comparable to the AAC developed by Vrba (1975) for bovids, this index has been used to document environmental change in the micromammal fossil record in eastern Africa (Jaeger,

1976; Wesselman, 1985; Fernandez-Jalvo et al., 1998; Reed, 2003, 2007; Reed and Gerads,

2011). In addition, Reed (2003) found correlations between Dendromurinae:Murinae and

Soricidae:Murinae to several ecological variables in his work on modern small mammal assemblages in the Serengeti ecosystem. All three ratios are simply calculated by dividing the number of individuals (MNI) in the taxon of interest (i.e. Gerbillinae) by the number of murines.

The membership of Otomys in the sub-family Murinae has been confirmed (Happold, 2013), but owing to its highly derived laminate dentition, it was not recognized as a murine in previous publications and thus was excluded from taxonomic ratio calculations. Ratio data has been calculated in two ways, both including and excluding Otomys from the murine category. In general, all three ratios operate on the premise that the murinae occupy more mesic environments than the taxon to which they are being compared.

6.2.4 Correspondence Analysis

Correspondence analyses (CA) is a multivariate ordination technique that maps data to a reduced number of dimensions making it easier to visualize and interpret (Hammer and Harper,

2006). Correspondence analyses (CA) have been used to plot taxa and fossil sites together so that 131 trends and groupings can be more easily visualized. Plots were generated using taxon presence/absence and relative abundance data. Due to the high diversity of small mammal taxa included in these analyses, taxon names have been removed from all plots for clarity. However, patterns of association between taxa and sites are discussed in the text. De-trended correspondence analyses (DCA) have been used in some cases (indicated for relevant figures) to reduce compression of data points which occurs in correspondence analyses (for details regarding detrending and rescaling for DCA, see Hammer and Harper, 2006). Most of the ordination plots include Afrosoricida, Macroscelidea, Rodentia, and Soricomorpha and use relative abundance data, unless otherwise indicated.

6.2.5 Taxonomic Habitat Indices

A Taxonomic Habitat Index (THI) is a cumulative index obtained by combining the habitat indications of all species contained in an assemblage (Nesbit-Evans et al., 1981;

Fernandez-Jalvo et al., 1998; Reed, 2003, 2007). The ecological preference of taxa is based on the habitat in which extant species live. For each taxon, a score is allocated to various types of pre-established vegetation in which the species are found. The ecological preferences of extant small mammal taxa were allocated using information from the available literature on the autoecology of all taxa in all assemblages. In essence, THI analyss aggregate information for each micromammal taxa present in an assemblage into a composite representation of the paleoenvironment (Nel and Henshilwood, 2016). The cumulative index should indicate the dominant habitats characterizing each assemblage. The reader is referred to Chapter 5 (and

Table A.1) for detailed information about the THI employed in this analysis. In this chapter habitat proportions were estimated using all taxa (Afrosoricida, Macroscelidea, Rodentia, and

Soricomopha), weighted for the relative abundance were used (unless otherwise indicated). 132 Stacked histograms (referred to as habitat spectra), indicating the relative proportions of each habitat category predicted for each assemblage were generated based on the THI for each fossil assemblage.

All statistical calculations were made using the free software program for analyses of paleontological data; Paleontological Statistics v3.1 (PAST). Figures were generated using PAST

(Version 3.1) and Microsoft Excel (Version 16.9).

RESULTS AND DISCUSSION

In all, the southern African small mammal fossil assemblages included in this study comprise 29,313 specimens. Sample sizes range from 157 individuals (MNI) at ST-M4 to 6,265 at SK-M1. Taxonomic richness ranges from 17 (ST-M4) to 27 (KB) genera. Certain patterns in the data are evident. For example, a single taxon, Mystromys dominates most of the Plio-

Pleistocene assemblages to a significant degree. Otomys and Sengi’s (elephant shrews) are also strongly represented. These abundance data clearly contrast with abundance data from modern barn owl roosts in the area (see Table A.4, A.5; personal observation). Notably, Mystromys albicaudatus is considered locally extinct in the Sterkfontein Valley today and has been designated as “Endangered” according to the IUCN redlist (Coetzee and Monadjem, 2008 IUCN

Red List).

6.3.1 Rarefaction

Rarefaction curves were generated for all fossil sites and are illustrated in Figure 6.2 (for rarefaction curves including eastern African fossil sites, see Figure A.1 and Figure A.2 in the appendix). Outer curves display 95% confidence intervals. The curves for each site sampled are somewhat difficult to see because of the number of sites overlain on the figure, however, it is 133 clear that the curves for most of the sites level off. This suggests that the samples sizes for these assemblages adequately reflect richness. Only two curves, ST-M4 and KA (Figure 6.3), do not appear to level off. Richness appears to be higher at KB and WW. Interestingly, the curves for

GON and DR level off despite relatively small sample sizes at both sites. When samples were rarified and standardized to n=157 (ST-M4), no statistical differences were found in richness (see

Table 6.2 for results of statistical tests). It does not appear, therefore, that richness is under- represented at ST-M4 despite its smaller sample size.

Figure 6.2: Individual rarefaction curves for all southern African fossil sites with estimated species richness and 95% confidence interval.

134

Figure 6.3: Individual rarefaction curves for ST-M4 and KA with estimated species richness and 95% confidence interval.

Table 6.2: Standardization of individual rarefaction curves Species Standard Permutation Site Code Mean t-test p-value Richness Deviation t-test STS-M4 17.1641 0.846019 11.59 ------MAK 17.4207 1.13669 13.72 1.2554 0.2197 0.2247 ST-M5 15.4666 1.50771 11.42 1.042 0.9178 0.9184 SK-M1 14.8336 1.63198 10.57 0.6451 0.5241 0.772 SK-M2 15.7725 1.52822 11.33 0.1592 0.8747 0.8779 SK-M3 13.6555 1.50959 9.853 1.1378 0.2649 0.2634 KB 17.521 1.73765 12.26 -0.3859 0.7025 0.7031 KA 14.3225 1.3809 10.34 0.0803 0.4286 0.4276 DR 17.1884 1.05878 13.18 -0.9497 0.3504 0.3602 GON 18.2696 1.07073 14.098 -1.4516 0.1577 0.1606 WW-S12 19.6377 1.35016 15.097 -1.9419 0.0623 0.0621 WW-S11 18.4131 1.46008 14.103 -1.4473 0.1589 0.1641 GVED 19.0452 1.42712 13.98 -1.3387 0.1914 0.1912

6.3.2 Shannon-Weiner Index, Evenness, Simpson’s Dominance

Several measures of diversity have been calculated for the southern African sites and are presented in Table 6.3. However, only the Shannon-Weiner Index and the Simpson index of 135 dominance are discussed below. Diversity indices for BF, LBW, and TAUNG are not given because presence/absence data were the only data available for these sites at the time of analysis.

No clear trend in diversity is apparent from older to younger sites (Figure 6.4). Of the hominin-bearing assemblages, SK-M3 is the least diverse (H’=1.688), followed by ST-M4, and the remaining Swartkrans deposits. This is not unanticipated given the very high abundance of the genus Mystromys in these assemblages and the relatively small sample size at ST-M4.

Although the Swartkrans and Sterkfontein samples are large and species richness is high, the dominance of Mystromys depresses overall diversity. The effect of this taxon is also clearly evident in the low Evenness and high Simpson’s index of dominance for both sites. The greatest diversity is seen at GON (H’=2.55), followed by MAK and DR

. 136

Shannon Weiner Diversity Index (H') 2.7

2.5

2.3

H' 2.1

1.9

1.7

1.5

Simpson's Index of Dominance (D) 0.35

0.3

0.25

0.2 D 0.15

0.1

0.05

0 MAK ST-M4 ST-M5 SK-M1 DR GON KA KB SK-M2 SK-M3 GVED 3.0-2.6 Ma 2.6 - 2 Ma 2.0-1.5 Ma 2.0 - 1.8 Ma 2.0 - 1.4 Ma 1.95 - 1.78 Ma 1.9 - 1.5 Ma 1.8 - 1.6 Ma 1.7 - 1.1 Ma 1.3 - 0.6 Ma 0.780 - 0.560 Ma

Figure 6.4: Shannon Weiner Diversity Index (H’) and Simpson’s Index of Dominance (D) for Makapansgat and all Sterkfontein Valley fossil sites. Fossil sites are arranged in chronological order from oldest (left) to youngest (right).

137 Results of statistical comparisons of the Shannon-Weiner index for general diversity and the Simpson index for dominance for all sites (using a standard t-test where p<0.05) are presented in Table 6.4. In general, an overall pattern ermerges. The Sterkfontein, Swartkrans, and Kromdraai deposits are all characterized by low diversity and, in many cases, do not differ significantly from one another. Similarly, Makapansgat, Gondolin, and Drimolen are relatively higher in diversity and do not consistently differ from one another. 138

Table 6.3: Diversity Indices Summary

Site Taxa S MNI Dominance D Simpson 1-D Shannon H Evenness Brillouin Menhinick Margalef Equitability J Fisher alpha Berger-Parker Chao-1

MAK 22 4358 0.1029 0.8971 2.55 0.5822 2.535 0.3333 2.506 0.825 3.025 0.1902 22

ST-M4 17 156 0.3244 0.6756 1.709 0.3069 1.559 1.441 3.366 0.5913 5.258 0.5321 27.33

ST-M5 23 4704 0.2693 0.7307 1.938 0.302 1.925 0.3353 2.602 0.6181 3.146 0.487 23

SK-M1 23 6265 0.3239 0.6761 1.749 0.25 1.739 0.2906 2.516 0.5578 3.01 0.5419 23

SK-M2 23 2474 0.2522 0.7478 1.941 0.3029 1.919 0.4624 2.816 0.6191 3.506 0.458 23

SK-M3 23 3002 0.3303 0.6697 1.688 0.2353 1.671 0.4198 2.748 0.5385 3.388 0.5453 26

KB 27 1982 0.2841 0.7159 1.925 0.2539 1.895 0.6065 3.425 0.5841 4.421 0.5005 28

KA 18 368 0.3033 0.6967 1.751 0.3199 1.67 0.9383 2.877 0.6057 3.963 0.5136 18.75

DR 19 438 0.1102 0.8898 2.45 0.6099 2.364 0.9079 2.959 0.8321 4.049 0.1758 19

GON 20 254 0.09945 0.9006 2.555 0.6439 2.418 1.255 3.431 0.853 5.089 0.2047 20.2

WW-S12 26 3505 0.0829 0.9171 2.706 0.5757 2.686 0.4392 3.063 0.8305 3.809 0.1549 26

WW-S11 26 1660 0.09504 0.905 2.595 0.5151 2.559 0.6381 3.372 0.7964 4.376 0.1723 26.33

GVED 23 436 0.1284 0.8716 2.425 0.4916 2.329 1.102 3.62 0.7736 5.173 0.2546 26

ARA 12 1019 0.285 0.715 1.682 0.4479 1.655 0.3759 1.588 0.6768 1.91 0.4897 12

FLKN1-6 14 424 0.1199 0.8801 2.287 0.7036 2.219 0.6799 2.149 0.8668 2.781 0.2099 14

FLKNN1-3 14 117 0.1175 0.8825 2.315 0.7236 2.129 1.294 2.73 0.8774 4.149 0.1966 14.5

HADAR 8 1246 0.2108 0.7892 1.651 0.6513 1.635 0.2266 0.9821 0.7938 1.144 0.2624 8

OMO B 11 126 0.4153 0.5847 1.4 0.3688 1.281 0.98 2.068 0.584 2.899 0.627 17

139

Table 6.4: Results of t-test for Shannon Wiener and Simpson Indices p (same) Shannon Wiener Index Site Code MAK ST-M4 ST-M5 SK-M1 SK-M2 SK-M3 KB KA DR GON WW-S12 WW-S11 GVED MAK *** *** *** *** *** *** *** * 0.0368 NS *** *0.0277 *0.0273 Simpson Index STS-M4 *** NS NS NS NS NS NS *** *** *** *** *** ST-M5 *** NS *** NS *** NS ** 0.0097 *** *** *** *** *** SK-M1 *** NS *** *** NS *** NS *** *** *** *** *** SK-M2 *** NS * 0.0843 *** *** NS * 0.010 *** *** *** *** *** SK-M3 *** NS *** NS *** *** NS *** *** *** *** *** KB *** NS NS *** * 0.011 *** * 0.0238 *** *** *** *** *** KA *** NS NS NS * 0.035 NS NS *** *** *** *** *** DR NS *** *** *** *** *** *** *** NS *** **0.0017 NS GON NS *** *** *** *** *** *** *** NS **0.0051 NS NS WW-S12 *** *** *** *** *** *** *** *** *** * 0.039 *** *** WW-S11 **0.0038 *** *** *** *** *** *** *** * 0.0121 NS *** ** 0.0019 GVED **0.0056 *** *** *** *** *** *** *** NS * 0.0119 *** ***

Significance * p < 0.05 ** p < 0.01 *** p < 0.0001 140

6.3.3 Similarity Indices

Similarity in taxonomic composition between assemblages was compared on a high-level

(presence/absence) and low level (relative abundance). Four eastern African assemblages have been included for comparison. In the dendrogram based on presence/absence data (Figure 6.5), the eastern African sites are clear outgroups. BF, LBW, and WW differ from the rest of the southern African sites. MAK, ST-M4, and GVED all separate from an ‘inner group of sites, which, (with the exception of Taung) are Paranthropus robustus sites. Within this inner grouping, DR and GON form their own cluster, as do Taung, ST-M5, and KB. The SK deposits all cluster tightly. 141

Figure 6.5: Dendrogram based on Jaccard similarity index evaluating species composition based on presence/absence data. The results were bootstrapped (n=9999). Sites associated with Australopithecus africanus are indicated in blue. Sites associated with Paranthropus robustus are indicated in red.

The dendrogram generated using relative abundance data (Figure 6.6) is similar but not identical. Instead, MAK groups with DR and GON and ST-M4 groups with ST-M5.

Interestingly, KB does not group with KA but instead consistently groups with the Swartkrans deposits. 142

Figure 6.6: Dendrogram based on Bray-Curtis similarity index evaluating species composition based on relative abundance data. The results were bootstrapped (n=9999). Sites associated with Australopithecus africanus are indicated in blue. Sites associated with Paranthropus robustus are indicated in red.

6.3.4 Taxonomic Ratios

Taxonomic ratios are presented in Table 6.5 and plotted in Figure 6.7. Sites are arranged chronologically. Taken together, no clear congruity exists between the three ratios – at some sites one ratio will increase and the other will decrease. The G:M ratios at ST-M5, SK-M1, SK-M3, and KB is slightly higher than at MAK, DR, or GON. However, any conclusions drawn from 143 these ratios would be premature and tenuous at best. Overall, there is little evidence of an increase in the G:M ratio over time.

Taxonomic Ratios 1.4

1.2

1

0.8 G:M D:M S:M 0.6

0.4

0.2

0 MAK ST-M4 ST-M5 SK-M1 DR GON KA KB SK-M2 SK-M3 GVED

Figure 6.7: Taxonomic ratios including Gerbillinae:Murinae, Dendromurinae:Murinae, Soricidae:Murinae. A value of 1 indicates equal proportions of each taxon. Fossil sites are arranged in chronological order from oldest (left) to youngest (right).

144

Table 6.5: Taxonomic ratios excluding and including the genus Otomys G:M D:M S:M Site Code G:M D:M S:M (Incl. OTOM) (Incl. OTOM) (Incl. OTOM) MAK 0.007 0.443 1.079 0.003 0.211 0.513 ST-M4 0.053 1.316 0.000 0.037 0.926 0.000 ST-M5 0.369 0.768 0.489 0.158 0.329 0.210 SK-M1 0.170 0.153 0.246 0.067 0.060 0.096 SK-M2 0.073 0.188 0.192 0.033 0.084 0.086 SK-M3 0.159 0.084 0.159 0.059 0.031 0.059 KB 0.293 0.571 0.211 0.108 0.212 0.078 KA 0.053 1.316 0.000 0.037 0.926 0.000 DR 0.010 0.434 0.626 0.006 0.244 0.352 GON 0.026 0.564 0.487 0.021 0.454 0.392 WW-S12 1.796 0.778 0.931 1.373 0.595 0.711 WW-S11 1.706 0.576 0.604 1.056 0.357 0.374 GVED 0.111 0.313 0.263 0.066 0.186 0.156 ARA 0.166 0.000 0.248 0.166 0.000 0.248 FLKN1-6 1.098 0.696 0.000 0.586 0.372 0.000 FLKNN1-3 0.224 0.254 0.000 0.181 0.205 0.000 HADAR 0.084 0.000 0.000 0.084 0.000 0.000 OMO B 0.064 0.000 0.083 0.064 0.000 0.083

6.3.5 Correspondence Analysis

All small mammal fossil assemblages (eastern African included) have been plotted with the biozones and modern roosts from Chapter 5 in Figure 6.8. The vertical axis of the plot separates more arid environments from more mesic ones while the horizontal axis separates east

African assemblages from southern ones. All of the southern African fossil sites plot near the

KAL, HIG, and CAPE. The eastern African fossil sites are more scattered, but generally plot near eastern African biozones, with the Olduvai deposits (FLKNN1-3 and FLKN1-6) nearest to the southern African sites. 145

Figure 6.8: Detrended correspondence analyses based on presence/absence data and including all biozones, roost sites, and fossil sites. Eastern African sites and biozones distributed on the right side of the plot and southern African biozones, roosts, and fossil sites are distributed on the left. More arid biozones and roosts plot in the upper half of the DCA while more mesic habitats plot in the lower half.

In Figure 6.9 the southern African fossil sites are plotted (DCA) with the modern roost sites from Chapter 5 using presence/absence data. Fossil and modern sites separate along the horizontal axis. Though there is significant taxonomic overlap between modern and fossil faunas, this separation is driven by the presence of extinct taxa in the fossil assemblages. With the exception of BF, LBW and MAK, most of the southern African fossil sites group most closely 146 with modern roosts nearest to the present day Sterkfontein Valley. Although sites were plotted using presence/absence the pattern persists when relative abundance data are used (no figure shown).

Figure 6.9: Detrended correspondence analyses based on presence/absence data and including all southern African roost and fossil sites. Modern roosts are distributed on the right side of the plot and fossil sites are distributed on the left. Roosts in more arid habitats plot in the upper half of the DCA while more mesic habitats plot in the lower half.

147 In Figure 6.10 the southern African fossil sites are plotted together using relative abundance. Several observations can be made from this plot. First, ST-M4 and 5, SK-M1-3, and

KB all cluster closely together while the WW deposits, GVED, GON, DR, and MAK all plot separately. The ordination of WW, in the upper left quadrant of the plot, makes sense as it is located far to the west of the Sterkfontein Valley sites, in the dry central portion of southern

Africa. MAK, GON, and DR all occur in the lower right quadrant of the plot but are not closely grouped like the Sterkfontein and Swartkrans sites. This basic arrangement does not change when analyses are limited to rodents only, weighted for relative abundance.

148

1.0

0.5 SK-M3 SK-M1 SK-M2 KB

ST-M5 ST-M4 KA WW-S11 WW-S12 -2.0 -1.5 -1.0 -0.5 GVED 0.5 1.0

-0.5

GON

DR

-1.0

MAK

-1.5

-2.0

Figure 6.10: Correspondence analysis including southern African fossil sites only based on all taxa and weighted for relative abundance. WW deposits, GON, DR, MAK, and GVED plot separately from the rest of the fossil sites.

6.3.6 Habitat Reconstructions using THI

6.3.6.1 Modern Biozones and Fossil Sites

Percentage predicted woody vs non-woody cover, based on presence/absence in modern biozone and fossil communities is presented in Figure 6.11 for context. All southern African sites are positioned towards the middle of the figure, between the HIG and NAM biozones. woody vegetation for all southern African fossil sites is indicated at between 36-54%. Only the 149 eastern African deposits at Hadar and Olduvai exceed this range for woody vegetation. In general, the fossil sites fall well within the “mixed” woody and non-woody area of the figure, which is expected given that most hominin sites were believed to be highly mosaic in nature

(Reynolds et al., 2015).

100%

90%

80%

70%

60%

50%

Non-Woody 40% Woody

30%

20%

10%

0% 6 3 - - BF SH M5 M4 M1 M2 M3 KB DR S11 KA S12 ------EFS - GUI SEL HIG - SUD ZBZ SAH FOR ARA KAL MAS GON LBW NAM MON MAK CAPE GVED ST ST SK SK SK OMO B TAUNG WW WW FLKN1 FLKNN1

HADAR

Figure 6.11: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for all biozones and all fossil sites based on the THI for all taxa, using presence/absence data. Sites are organized in ascending order from left to right by percentage non-woody vegetation.

6.3.6.2 Previously Studied Sites

Habitat spectra for woody vs. non-woody habitat categories, this time weighted for relative abundance are show in Figure 6.12. Only Makapansgat, Sterkfontein, Swartkrans, and 150 Kromdraai are included in the first two spectra. These sites were selected because they represent the “classic” southern African hominin sites which feature prominently in the paleoenvironmental reconstructions under consideration here (Vrba, 1975, 1985; Reed, 1997).

Fossil sites are organized chronologically in all figures. The highest percentage woody vegetation is indicated at MAK (45%). Percentage woody vegetation is estimated at between 30-

37% at all other sites. ST-M4 falls on the lower end of this range (~33%). A slightly higher proportion woody vegetation is indicated for all SK deposits and for ST-M5 in comparison to

ST-M4, although the differences are small.

100%

90%

80%

70%

60%

50% Non-Woody Woody

40%

30%

20%

10%

0% MAK ST-M4 ST-M5 SK-M1 KB KA SK-M2 SK-M3

Figure 6.12: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for previously studied fossil sites based on the THI for all taxa,weighted for relative abundance. Sites are organized chronologically from left to right.

151 Figure 6.13 shows habitat spectra generated using the five habitat categories. The dominant habitat category indicated at all sites is grassland (46-65%). Forest and woodland are highest at MAK (together ~23%), while at all other sites (except SK-M2) these habitat categories are estimated at < 10%. ST-M4 deposit does not differ appreciably from ST-M5, SK, or KB.

100%

90%

80%

70%

60% Semi-Arid Grassland 50% Bushland Woodland

40% Forest

30%

20%

10%

0% MAK ST-M4 ST-M5 SK-M1 KB KA SK-M2 SK-M3

Figure 6.13: Habitat spectra with all five habitat categories for previously studied fossil sites based on the THI for all taxa, weighted for relative abundance. Sites are organized chronologically from left to right.

6.3.6.3 All Sterkfontein Valley Fossil Sites

For Figure 6.14 and Figure 6.15 DR, GON, and GVED have been added. It is evident from both figures that none of the additional sites are similar in predicted habitat composition to

ST, SK, and KB. Instead, higher percentages woody vegetation (44-45%) are indicated at all 152 three sites, comparable to the proportions of these habitat categories for MAK. In the five habitat category spectra, forest and woodland categories for DR and GON are estimated at 22-23% of

(Gladysvale is slightly lower at 16%), compared with < 10% at ST, SK, and KB/KA. Grassland remains the dominant habitat category indicated for all habitat sites (46-65%).

When the THI was restricted to rodents only, weighted for relative abundance for the southern African fossil sites, forest, woodland, and semi-arid categories all decrease while grassland and bushland categories increase slightly (~ 5-10%). However, the overall composition of the sites and differences between them remain unchanged. Higher proportions of forest and woodland are indicated for MAK, DR, GON, and GVED compared to ST, SK, and KB/KA.

153

100%

90%

80%

70%

60%

50% Non-Woody Woody

40%

30%

20%

10%

0% MAK ST-M4 ST-M5 SK-M1 DR GON KB KA SK-M2 SK-M3 GVED

Figure 6.14: Habitat spectra with dichotomous representation of woody versus non-woody vegetation including DR, GON, and GVED. Spectra are based on the THI for all taxa, weighted for relative abundance. Sites are organized chronologically from left to right.

154

100%

90%

80%

70%

60% Semi-Arid Grassland 50% Bushland Woodland

40% Forest

30%

20%

10%

0% MAK ST-M4 ST-M5 SK-M1 DR GON KB KA SK-M2 SK-M3 GVED

Figure 6.15: Habitat spectra with all five habitat categories including DR, GON, and GVED. Spectra are based on the THI for all taxa, weighted for relative abundance. Sites are organized chronologically from left to right.

6.3.6.4 Eastern and southern African Fossil Sites

As a final, broad comparison, habitat spectra for eastern and southern African micromammal assemblages are presented (Figure 6.16 and Figure 6.17). In this instance, rodents only, weighted for relative abundance were used [because this is the level of analyses used by

Fernandez-Jalvo et al. (1998) and Reed (2003)]. A significant grassland component is indicated at the Ardipithecus ramidus site Aramis, while Hadar is reconstructed to have the greatest amount of woody vegetation. The Omo and Olduvai deposits are more wooded than the south

African sites, but these differences are not large. Most strikingly, the eastern African sites are 155 estimated to have more woody vegetation than the southern African sites for which the dominant habitat category is grassland.

100%

90%

80%

70%

60%

50% Non-Woody Woody 40%

30%

20%

10%

0%

Figure 6.16: Habitat spectra with dichotomous representation of woody versus non-woody vegetation for eastern and southern African fossil sites. Spectra are based on the THI for rodents only, weighted for relative abundance. Eastern African sites are on the left side of the figure and southern African sites are on the right. Sites are organized chronologically from left to right.

156

100%

90%

80%

70%

60%

Semi-Arid 50% Grassland Bushland 40% Woodland Forest

30%

20%

10%

0%

Figure 6.17: Habitat spectra with all five habitat categories for eastern and southern African fossil sites. Spectra are based on the THI for rodents only, weighted for relative abundance. Eastern African sites are on the left side of the figure and southern African sites are on the right. Sites are organized chronologically from left to right.

DISCUSSION

6.4.1 Swartkrans and Sterkfontein

In general, these results of the analyses agree with Avery’s (2001) assessment of the micromammalian assemblages at Sterkfontein and Swartrkans. The deposits from these two sites have similar biodiversity index values, group together according in both cluster and correspondence analyses, and do not differ greatly in their expected habitat proportions. There is therefore little evidence for significant difference in taxonomic or habitat composition between 157 the sites. With regards to the specific question of whether or not fundamental differences are indicated to exist between Sterkfontein and Swartkrans, the micromammal record is not in agreement with the large mammal record. Moreover, higher proportions of grass and semi-arid habitat are indicated for these fossil sites in the Plio-Pleistocene than are estimated for modern owl roosts located in the same area (see Figure 5.10, Chapter 5) and provide tentative support for

Avery’s argument that climatic conditions were hotter and drier in the area than they are today.

6.4.2 Environmental change through time

No consistent pattern of environmental change through time was detected in the micromammal record. MAK, GON, DR, and GVED are all characterized by higher diversity than ST-M4, ST-M5, SK M1-3, KB, and KA. The sites include both the oldest hominin-bearing and youngest sites and thus span the 3 – 1 Ma timeframe under consideration here. No consistent pattern of increase was evident in the Gerbillinae:Murinae ratio which would suggest increased aridity. In ordination plots, younger and older sites did not always plot separately but instead the composition of micromammal communities overlapped significantly. Habitat reconstructions for

MAK, GON, DR, and GVED all predict relatively greater proportions of woody cover, particularly forest and woodland habitats, for these four sites.

6.4.3 Hominin habitat associations

6.4.3.1 Australopithecus africanus

Australopithecus africanus sites include MAK, ST-M4, and TAUNG. At this stage, relatively little can be said about TAUNG because only presence/absence data are available for this site. However, TAUNG groups with the rest of the Sterkfontein Valley sites in both cluster and correspondence analyses (see Figure 6.5 and Figure 6.9), suggesting that despite its 158 geographic distance from the other sites (see Figure 6.1), its micromammal taxonomic composition is similar.

Greater richness, diversity, and evenness were indicated for MAK in all cases, probably indicating greater habitat heterogeneity and/or mesic conditions for this site. In the ordination plots (Figure 6.9 and Figure 6.10), MAK differs consistently from most of the Sterkfontein

Valley sites, both in terms of presence/absence of taxa and relative abundance. The possibility that these differences are attributable to the geographic location of Makapansgat (see Figure 6.1) cannot be ruled out. However, the habitat composition suggested by micromammals in these analyses agrees well with reconstructions of this site in other studies (Rayner et al., 1993; Reed,

1997).

In contrast, the habitat composition predicted for ST-M4 deviates from most reconstructions for this deposit which interpret its paleoenvironment as relatively more closed and mesic than ST-M5, SK M1-3, and KB (Vrba, 1974, 1975, 1985; Reed, 1997; Lee-Thorp et al., 2007). In fact, in this habitat reconstruction ST-M4 is characterized by slightly lower proportions of woody vegetation than the Paranthropus robustus deposits ST-M5, SK, and KB.

Moreover, ST-M4 is not similar to the other Australopithecus africanus site, MAK in any significant way.

6.4.3.2 Paranthropus robustus

Paranthropus robustus deposits include ST-M5, SK M1-3, KB, DR, and GON. ST-M5,

SK M1-3, and KB are similar in nearly all respects. However, it is apparent from these analyses that DR and GON differ from the rest of the Paranthropus robustus deposits. DR and GON are higher in diversity and group separately from ST-M5, SK M1-3, and KB in cluster and correspondence analyses. Reconstructions using the THI consistently predict higher proportions 159 of woody vegetation, particularly in the forest and woodland categories for these two sites. In fact, DR and GON are more similar to MAK than to any other deposits.

Overall, the data do not support the argument that Paranthropus robustus associated primarily with more open/grassy environments. These findings are consistent with the results of stable carbon isotope, microwear, and some faunal analyses (de Ruiter et al., 2008; Scott et al.,

2005; Sponheimer et al., 2006) which interpret both taxa as habitat generalists.

6.4.3.3 Comparisons between eastern and southern micromammal assemblages

Preliminary comparisons between eastern African and southern African micromammal assemblages indicate that the two regions differ in their proportions of woody versus non-woody cover. Overall, more woody vegetation is indicated at the eastern African sites. Specifically, more forest and woodland habitat is predicted at Hadar, Omo Member B, and Olduvai

(FLKNN1-3 and FLKN1-6).

The exception to this general pattern in is Aramis, a ~ 4.4 Ma old site associated with

Ardipithecus ramidus, where a significant grassland component is predicted. This reconstruction conflicts with the “mesic woodland” proposed by Louchart et al. (2009) in their analyses of the small mammal fauna at this site. Upon, closer inspection, the high predicted grassland proportion is driven by Uranomys which makes up 44% of the relative abundance of small mammal taxa at

Aramis. Uranomys is scored as a grassland specialist and thus its dominance at Aramis significantly influences the predictions of the THI. Louchart et al. (2009) attribute the high relative abundance of Uranomys to “predator bias” introduced by barn owls foraging in the open grassland components of a wooded landscape and instead heavily emphasize woodland (as indicated by “wooded-habitat” mammalian and avian taxa) (pp. 66e3) in their reconstruction.

However, Levin et al. (2015) several early to mid-Pliocene hominin deposits, including Aramis, 160 and found that the d13C composition of soil carbonates at Aramis were higher than those of all other sites in the study. These findings indicate relatively high proportions of C4 vegetation at

Aramis and support the habitat reconstruction in this study.

Reconstructions based on micromammal data for the other eastern African sites do not contradict most previously published paleoenvironmental interpretations of the represented deposits at Hadar, Omo, and Olduvai. For instance, the Sidi Hakoma (3.42 – 3.24 Ma) deposit at

Hadar (associated with Australopithecus afarensis) is reconstructed by Reed (1997) as a medium to open density woodland. The microfauna indicate a significant proportion of woody vegetation consisting of roughly equal proportions forest and woodland for this deposit. Similarly, Omo

Member B has been variably reconstructed a closed woodland/riverine forest (Jaeger and

Wesselman, 1976; Bonnefille, 1984; Wesselman, 1984; Reed, 1997) which is supported by the habitat reconstruction obtained here. Finally, the pattern of increased aridity at Olduvai from

FLKNN1-3 to FLKN1-6 and documented by both Fernandez-Jalvo et al., (1998) and Reed

(2003, 2007) is clearly replicated regardless of the modifications which were made to the niche models of taxa for this study. Interestingly, a significant semi-arid component is indicated at many of the eastern African sites (e.g. Hadar). It is possible that the micromammal assemblages at these sites derive from riparian woodland and forest habitats situated in environments that are overall more arid.

CONCLUSION

In conclusion, these analyses support Avery’s (2001) argument that Sterkfontein and

Swartkrans did not differ significantly in climate and vegetation. The micromammal record does not record any evidence of a clear trend towards the dominance of increasingly open, grassland habitats in the Sterkfontein Valley during the Plio-Pleistocene. Open habitats are strongly 161 indicated at Sterkfontein, Swartkrans, and Kromdraai while relatively more wooded habitats are predicted for Makapansgat, Drimolen. Gondolin, and Gladysvale. Finally, no consistent pattern of habitat association is indicated for Australopithecus africanus or Paranthropus robustus.

Instead, this study agrees with interpretations which categorize these taxa as habitat generalists.

Overall, the results of these analyses indicate that the environment in southern Africa was quite open between 3 and 1 Ma, but that patches of more mesic C3 woodland habitat were present.

Australopithecus africanus and Paranthropus robustus appear to have associated with both habitat types, perhaps taking advantage of resource-rich C3 habitat patches whenever they were available. 162 7 CONCLUSION

A primary goal of this dissertation was to determine whether patterns of evolutionary change observed in the large mammal fossil record of southern Africa are also evident in the small mammal record. Large mammals and other proxies (e.g. stable isotope analyses) indicate that a significant environmental transition characterized by a decrease in more mesic, C3 woodland habitats and expansion of more xeric, open C4 grassland habitats, took place between 3 and 1 Ma (Vrba, 1975, 1985; Reed, 1997). These changes are documented in multiple lineages, but especially in bovids, by decreased proportions of closed-habitat adapted taxa and increased proportions of open-adapted taxa (Vrba, 1974, 1975, 1985, 1995). A similar evolutionary response may have occurred in early hominins. Approximately 2 Ma, Australopithecus africanus, which is argued to have occupied relatively more wooded habitats, went extinct and was replaced in the region by Paranthropus robustus and early representatives of the genus

Homo.

However, previous analyses of the small mammal fossil record in the region found little evidence of change in micromammalian community composition between 3 and 1 Ma (Avery,

2001). Investigating this apparent contradiction between the large and small mammal record was the focus of the current study. A combination of stable carbon isotope analyses and taxon- dependent analyses were deployed, and in some cases significantly developed, in re-assessing the micromammalian fossil record associated with early hominins.

Most stable isotope analyses have focused on large mammals and few studies have used small mammals. However, recent work has suggested that the isotopic compositions of small mammals have potential for recording dietary behavior, habitat composition, and environmental 163 change over short and long timescales (Gehler et al., 2012; Hynek et al., 2012; Robb et al., 2012;

Dammhahn et al., 2013; Kimura et al., 2013; Symes et al., 2013; Codron et al., 2015; Jeffrey et al., 2015; Smiley et al., 2016; Commendador and Finney, 2016). Though some studies have begun investigating small mammal stable isotope ecology (see Gehler et al., 2012; Hynek et al.,

2012; Jeffrey et al., 2015; Smiley et al., 2015), large gaps remain in our knowledge of small mammal isotopic ecology across spatial scales, seasons, and heterogeneous landscapes. This is especially true in African ecosystems, where C4 plants are abundant and many landscapes relevant to hominin evolution persist.

Chapters 2 and 3 focused on small mammal stable carbon isotope ecology in the

Sterkfontein Valley. The emphasis of both chapters is on the relationship between small mammal stable carbon isotope composition and habitat, with the goal of applying this information to the fossil record.

Chapter 2 focused specifically on the stable carbon isotope ecology of small mammal insectivores. Insectivores were selected because it was hypothesized that they might act as isotope ecological “integrators” and therefore serve as useful proxies for overall small mammal community d13C composition. In this chapter, the degree to which carbon isotope compositions of three sympatric shrew species recorded spatial and temporal changes in habitat conditions in a mosaic southern African savanna environment was assessed. Habitats ranged from very open (<

5% canopy cover) to wooded (~60% canopy cover). Shrew hair d13C values were compared between microhabitat types, and across taxa, in order to test whether these data followed predictable patterns based on local vegetation. Shrew carbon isotope values were found to vary with habitat in a predictable manner. While taxonomy influenced d13C values, this was largely due to differences in habitat preferences of individual taxa and resultant variation in their relative 164 abundance within each environment. Isotopic differences between habitat types were preserved within individual taxa where taxa occurred in multiple habitats. At least in this study, insectivore carbon isotope compositions tracked changes in the proportions of C3 and C4 vegetation in the local environment. The fact that insectivores clearly evince a habitat signal is promising, both for modern ecological studies and for application to the fossil record.

Chapter 3 addressed broader questions about rodent community d13C composition in specific habitats under taphonomically controlled. In this chapter the degree to which carbon isotope compositions of small mammal tooth enamel record spatial changes in habitat was assessed. Modern small mammal specimens were collected from the pellet accumulations of three barn owl (Tyto alba africanus) roosts located within habitats varying from open grassland to mixed woodland. Rodent carbon isotope compositions within taxa, between taxa, and between tooth types were analyzed in an effort to characterize variation within this group. Rodent

13 community d Cenamel composition between habitat types was also examined to evaluate how well small mammal carbon isotope data reflect vegetation composition local to the roosts. The results of these analyses suggest that the relationship between small mammal community carbon isotopic means and vegetation composition is complex, but that with appropriate taxonomic control and consideration of relative abundance, small mammals have potential as a proxy for reconstructing past habitats at fine scales.

Preliminary results of stable carbon isotope analyses of fossil small mammal tooth enamel are also presented in Chapters 2 and 3. Both insectivore and rodent fossils exhibit higher

13 d C values than their modern counterparts. This indicates a greater contribution of C4 resources

to the diets of small mammals and may reflect more C4 grass availability in the past than characterizes the area today. 165 Chapters 4, 5, and 6 all focus on the taxonomic composition of micromammalian assemblages from the Sterkfontein Valley deposits. In Chapter 4, over 1,000 new fossil micromammal specimens from four deposits; Kromdraai B, Sterkfontein Member 4, Swartkrans

Member 1 (Hanging Remnant), and the Gladysvale External Deposits are described. The taxonomic composition of these samples was not found to differ fundamentally from that which had been previously documented at each site and these new materials significantly augmented the available datasets for these deposits.

In Chapter 5, a taxonomic habitat index (THI) was developed for African small mammals. Taxonomic habitat indices (THI) are models which incorporate data on the habitat associations of all fauna within a given community in a systematic way to produce a composite representation of the environment from which the community derives (van Couvering, 1980;

Simpson, 1960; Nesbit-Evans et al., 1981; Fernandez-Jalvo et al., 1998; Reed, 2003, 2007; Nel and Henshilwood, 2016; Nel et al., 2017). The accuracy of this model in predicting habitat compositions was evaluated using small mammal community data from modern biozones and barn owl (Tyto alba) roosting sites throughout southern Africa. These analyses indicate that presence/absence data can be used to generate habitat reconstructions, but that this type of data is more likely predict the presence of habitat categories which do not actually exist.

Reconstructions based upon relative abundance data more accurately reflect habitat composition.

In all cases, the dominant habitat categories in each biozone and at each roost site were predicted. It was concluded, therefore, that this THI can be applied to the micromammalian fossil record with reasonable confidence that the dominant habitat elements will be represented in the paleohabitat reconstructions for each fossil deposit. 166 Finally, the paleoecological analyses of several sites from the Sterkfontein Valley were presented in Chapter 6. Several methods for assessing faunal communities including the THI developed in Chapter 5 were used. Three primary goals were established for this chapter; 1) to re-evaluate Avery’s (2001) argument that paleoenvironmental conditions in the Sterkfontein

Valley did not change significantly between 3 and 1 Ma, 2) to determine if micromammal fossil assemblages record an increase in the proportion of open, grassy habitats over time, and 3) to evaluate whether or not different habitats are indicated for Australopithecus africanus versus

Paranthropus robustus.

The results of these analyses support Avery’s (2001) argument that Sterkfontein and

Swartkrans did not differ significantly in climate or vegetation. Open, grassland habitats are strongly indicated by the THI for all deposits at Sterkfontein, Swartkrans, and Kromdraai. In addition, small mammal fossils from Sterkfontein Member 4 and Swartkrans Member 1 exhibit higher d13C values than their modern counterparts in the area. Therefore, stable carbon isotope data from small mammal fossil tooth enamel support the argument that all of these sites were mostly open, grassy habitats.

However, no clear trend towards increasingly open habitats is documented through time in this area. Although habitats at Sterkfontein, Swartkrans, and Kromdraai all appear quite open,

Drimolen, Gondolin, and Gladysvale, which are located less than 15 km away and are of similar age, are reconstructed as relatively more wooded. In fact, the habitat reconstructions for

Drimolen and Gondolin look similar to the reconstruction for Makapansgat. The micromammal data therefore suggest that the Sterkfontein Valley consisted of a mosaic landscape of open grassland with patches of moderately-closed woodland throughout the Plio-Pleistocene. 167 In addition, no consistent pattern of habitat association is indicated for either

Australopithecus africanus or Paranthropus robustus. The Australopithecus africanus deposits at

Sterkfontein Member 4 and Makapansgat are reconstructed as open grassland and moderately wooded respectively. Similarly, the Paranthropus robustus deposits Sterkfontein Member 5,

Swartkrans, and Kromdraai are reconstructed as open grassland while Drimolen and Gondolin are relatively more wooded. These findings are consistent with the results of stable carbon isotope, microwear, and some faunal analyses (de Ruiter et al., 2008; Scott et al., 2005;

Sponheimer et al., 2006) which interpret both taxa as habitat generalists.

It is also apparent from these analyses and other lines of evidence that there were significant regional differences between environments in eastern Africa and southern Africa during the Plio-Pleistocene. Overall, the results of these analyses indicate that the environment in southern Africa was quite open between 3 and 1 Ma, but that patches of more mesic C3 woodland habitat were present. Australopithecus africanus and Paranthropus robustus appear to have been associated with both habitat types, perhaps taking advantage of resource-rich C3 habitat patches whenever they were available.

168 8 BIBLIOGRAPHY

Abramsky, Z. Rosenzweig, M.L., 1984. Tilman's predicted productivity–diversity relationship shown by desert rodents. Nature 309(5964), p.150.

Abramsky, Z., 1978. Small mammal community ecology. Oecologia 34(2), 113-123.

Abramsky, Z., 1988. The role of habitat and productivity in structuring desert rodent communities. Oikos 107-114.

Acocks, J.P.H., 1988. Veld types of South Africa (No. 57, Ed. 3).

Adams, J.W., Herries, A.I.R., Conroy, G.C., Kuykendall, K.L., 2007. Taphonomy of a South African cave: geological and hydrological influences on the GD 1 fossil assemblage at Gondolin, a Plio-Pleistocene paleocave system in the Northwest Province, South Africa. Quaternary Science Review 2526-2543.

Altmann, S.A. and Altmann, J., 1973. Baboon ecology. University of Chicago Press.

Ambrose, S.H., DeNiro, M.J., 1986. The isotopic ecology of East African mammals. Oecologia 69(3), 395-406.

Andrews P., Evans E.N., 1983. Small mammal bone accumulations produced by mammalian carnivores. Paleobiology 289-307.

Andrews P., 1983. Small mammal faunal diversity at Olduvai Gorge, Tanzania. Animals and Archaeology 1, 77-85.

Andrews P., 2006. Taphonomic effects of faunal impoverishment and faunal mixing. Palaeogeography, Palaeoclimatology, Palaeoecology 241, 572-589.

Andrews, P., O'Brien, E.M., 2000. Climate, vegetation, and predictable gradients in mammal species richness in southern Africa. Journal of Zoology 251(2), 205-231.

Andrews, P., 1990. Owls, caves and fossils: predation, preservation and accumulation of small mammal bones in caves, with an analysis of the Pleistocene cave faunas from Westbury-sub- Mendip, Somerset, UK. University of Chicago Press.

Andrews, P., Lord, J.M., Evans, E.M.N., 1979. Patterns of ecological diversity in fossil and modern mammalian faunas. Biological Journal of the Linnean Society 11(2), 177-205.

Avenant, N., 2011. The potential utility of rodents and other small mammals as indicators of ecosystem 'integrity' of South African grasslands. Wildlife Research, 38(7), 626-639. 169 Avenant, N.L., Cavallini, P., 2007. Correlating rodent community structure with ecological integrity, Tussen-die-Riviere Nature Reserve, Free State province, South Africa. Integrative Zoology 2(4), 212-219.

Avenant, N.L., 2005. Barn owl pellets: a useful tool for monitoring small mammal communities. Belgian Journal of Zoology 135, 39-43.

Avery D., Vrba E., Denton G., Partridge T., Burckle L., 1995. Southern savannas and Pleistocene hominid adaptations: the micromammalian perspective. In: Vrba, E.S. ed., 1995. Paleoclimate and evolution, with emphasis on human origins. Yale University Press, pp. 459-478.

Avery D., 1992. Ecological data on micromammals collected by barn owls Tyto alba in the West Coast National Park, South Africa. Israel Journal of Zoology 38(3-4), 385-397.

Avery D., 1995. A preliminary assessment of the micromammalian remains from Gladysvale Cave, South Africa. Palaeontologia Africana 32, 1-10.

Avery, D.M., Avery, G., Palmer, N.G., 2005. Micromammalian distribution and abundance in the Western Cape Province, South Africa, as evidenced by Barn owls Tyto alba (Scopoli). Journal of Natural History 39(22), 2047 - 2071.

Avery, D.M., Stratford, D.J. and Sénégas, F., 2010. Micromammals and the formation of the Name Chamber at Sterkfontein, South Africa. Geobios 43(4), 379-387.

Avery, D.M., 1982. Micromammals as palaeoenvironmental indicators and an interpretation of the late Quaternary in the southern Cape Province, South Africa: South African Museum.

Avery, D.M., 1998. An assessment of the lower pleistocene micromammalian fauna from Swartkrans members 1–3, Gauteng, South Africa. Geobios 31, 393-414.

Avery, D.M., 2001. The Plio-Pleistocene vegetation and climate of Sterkfontein and Swartkrans, South Africa, based on micromammals. Journal of Human Evolution 41, 113-132.

Avery, D.M., 2003. Early and Middle Pleistocene environments and hominid biogeography; micromammalian evidence from Kabwe, Twin Rivers and Mumbwa Caves in central Zambia. Palaeogeography, Palaeoclimatology, Palaeoecology 189, 55-69.

Balter, V., Blichert-Toft, J., Braga, J., Telouk, P., Thackeray, F., Albarède, F., 2008. U–Pb dating of fossil enamel from the Swartkrans Pleistocene hominid site, South Africa. Earth and Planetary Science Letters 267(1-2), 236-246.

Bamford, M., 1999. Pliocene fossil woods from an early hominid cave deposit, Sterkfontein, South Africa. South African Journal of Science 95(5), 231-237. 170 Barton, R.A., 1990. Foraging strategies, diet and competition in olive baboons. Doctoral dissertation, University of St Andrews.

Barton, R.A., Whiten, A., Strum, S.C., Byrne, R.W., Simpson, A.J., 1992. Habitat use and resource availability in baboons. Animal Behaviour 43(5), 831-844.

Baugh, A.T., West, A.G., Rickart, E.A., Cerling, T.E., Ehleringer, J.R., Dearing, M.D., 2004. Stable isotope ratios (δ15N and δ13C) of syntopic shrews (Sorex). The Southwestern Naturalist 49(4), 493-500.

Bedaso, Z., Wynn, J.G., Alemseged, Z., Geraads, D., 2010. Paleoenvironmental reconstruction of the Asbole fauna (Busidima Formation, Afar, Ethiopia) using stable isotopes. Geobios, 43(2), 165-177.

Bedaso, Z.K., Wynn, J.G., Alemseged, Z., Geraads, D., 2013. Dietary and paleoenvironmental reconstruction using stable isotopes of herbivore tooth enamel from middle Pliocene Dikika, Ethiopia: Implication for Australopithecus afarensis habitat and food resources. Journal of Human Evolution 64(1), 21-38.

Behrensmeyer, A.K., Hill AP., 1988. Fossils in the making: vertebrate taphonomy and paleoecology: University of Chicago Press.

Behrensmeyer, A.K., Todd, N.E., Potts, R., McBrinn, G.E., 1997. Late Pliocene Faunal Turnover in the Turkana Basin, Kenya and Ethiopia. Science 278, 1589-1594.

Benton, M.J., Harper, D.A., 2009. Introduction to paleobiology and the fossil record: John Wiley & Sons.

Berger, L.R., 1993. A preliminary estimate of the age of the Gladysvale australopithecine site.

Berger, L.R., Keyser, A.W., Tobias, P.V., 1993. Gladysvale: first early hominid site discovered in South Africa since 1948. American Journal of Physical Anthropology 92(1), 107-111.

Bergstrom, B.J., 2013. Would East African savanna rodents inhibit woody encroachment? Evidence from stable isotopes and microhistological analysis of feces. Journal of Mammalogy 94(2), 436-447.

Bobe, R., Behrensmeyer, A.K., 2004. The expansion of grassland ecosystems in Africa in relation to mammalian evolution and the origin of the genus Homo. Palaeogeography, Palaeoclimatology, Palaeoecology, 207(3-4), 399-420.

Bobe, R., Behrensmeyer, A.K., Chapman, R.E., 2002. Faunal change, environmental variability and late Pliocene hominin evolution. Journal of Human Evolution 42(4), 475-497.

Bonnefille, R., 1984. Cenozoic vegetation and environments of early hominids in East Africa. The Evolution of the East Asian Environment 2, 579-612. 171 Bonnefille, R., Potts, R., Chalié, F., Jolly, D., Peyron, O., 2004. High-resolution vegetation and climate change associated with Pliocene Australopithecus afarensis. Proceedings of the National Academy of Sciences of the United States of America 101(33), 12125-12129.

Brachert, T.C., Brügmann, G.B., Mertz, D.F., Kullmer, O., Schrenk, F., Jacob, D.E., Ssemmanda, I., Taubald, H., 2010. Stable isotope variation in tooth enamel from Neogene hippopotamids: monitor of meso and global climate and rift dynamics on the Albertine Rift, Uganda. International Journal of Earth Sciences 99(7), 1663-1675.

Braga, J., Thackeray, J.F., Bruxelles, L., Dumoncel, J. and Fourvel, J.B., 2017. Stretching the time span of hominin evolution at Kromdraai (Gauteng, South Africa): Recent discoveries. Comptes Rendus Palevol 16(1), 58-70.

Brain, C.K., 1981. The hunters or the hunted?. University of Chicago Press.

Brain, C.K., 1993. A taphonomic overview of the Swartkrans fossil assemblages. Swartkrans: A Cave’s Chronicle of Early Man. Transvaal Museum, Pretoria, pp. 257-264.

Brain, V., CK, C., Clark, J.D., Grine, F.F., Shipman, P., Susman, R.L., Turner, A., 1988. New evidence of early hominids, their culture and environment from the Swartkrans cave, South Africa. South African Journal of Science, 84(10), 828.

Broom, R., Robinson, J.T., 1949. A new type of fossil man. Nature 164, 322-323.

Broom, R., 1908. Further observations on the Chrysochloridae. Annals of the Transvaal Museum 1, 14-16.

Broom, R., 1910. On Chrysochloris namaquensis. Transactions of the Royal Society of South Africa 2, 41-43.

Broom, R., 1939a. The fossil rodents of the limestone cave at Taungs. Annals of the Transvaal Museum 19, 315-317.

Broom, R., 1939b. A restoration of the Kromdraai skull. Annals of the Transvaal Museum 19, 327-329.

Broom, R., 1941. On two Pleistocene golden moles. Annals of the Transvaal Museum 20, 215- 216.

Broom, R., 1948. Some South African Pliocene and Pleistocene mammals. Annals of the Transvaal Museum 21, 1-38.

Broom, R., 1937. On some new Pleistocene mammals from limestone caves of the Transvaal. South African Journal of Science 33, 750-768. 172 Brown, J.S., Kotler, B.P., Smith, R.J., Wirtz, II W.O., 1988. The effects of owl predation on the foraging behavior of heteromyid rodents. Oecologia 76, 408-415.

Bruxelles, L., Maire, R., Beaudet, A., Couzens, R., Duranthon, F., Fourvel, J.B., Stratford, D., Thackeray, F., Braga, J., 2018. The revised stratigraphy of the hominin-bearing site of Kromdraai (Gauteng, South Africa) and associated perspectives. Journal of Human Evolution 114, 1-19.

Bump, J.K., Fox-Dobbs, K., Bada, J. L., Koch, P. L., Peterson, R. O., Vucetich, J. A., 2007. Stable isotopes, ecological integration and environmental change: wolves record atmospheric carbon isotope trend better than tree rings. Proceedings of the Royal Society of London B: Biological Sciences 274(1624), 2471-2480.

Bunn, D.S., Warburton, A.B., Wilson, R.D., 2010. The barn owl. A&C Black.

Butler, P.M., Greenwood, M., 1976. Elephant-shrews (Macroscelididae) from Olduvai and Makapansgat. Fossil vertebrates of Africa 4, 1-56.

Butler, P.M., Greenwood, M., 1979. Soricidae (Mammalia) from the Early Pleistocene of Olduvai Gorge, Tanzania. Zoological Journal of the Linnean Society 67(4), 329-379.

Butler, P.M., 1998. Fossil history of shrews in Africa. Evolution of shrews, pp.121-132.

Cartmill, M., 1967. The Early Pleistocene mammalian microfaunas of sub-Saharan Africa and their ecological significance. Quaternaria 9, 169-197.

Castiglia, R., Solano, E., Makundi, R.H., Hulselmans, J., Verheyen, E., Colangelo, P., 2012. Rapid chromosomal evolution in the mesic four-striped grass rat Rhabdomys dilectus (Rodentia, Muridae) revealed by mtDNA phylogeographic analysis. Journal of Zoological Systematics and Evolutionary Research 50(2), 165-172.

Cerling, T.E., 1992. Development of grasslands and savannas in East Africa during the Neogene. Palaeogeography, Palaeoclimatology, Palaeoecology 97, 241-247.

Cerling, T.E., 1984. The stable isotopic composition of modern soil carbonate and its relationship to climate. Earth and Planetary Science Letters 71(2), 229-240.

Cerling, T.E., 1999. Stable carbon isotopes in palaeosol carbonates. In: Thiry, M., Simon- Coinçon, R., (Eds.), Palaeoweathering, palaeosurfaces and Related Continental Deposits, 43-60.

Cerling, T.E., 2013. Stable isotope evidence for hominin environments in Africa. In: Holland, H., Turekian, K., (Eds.), Treatise on Geochemistry: Second Edition. Elsevier Inc.

Cerling, T.E., Harris, J. M., 1999. Carbon isotope fractionation between diet and bioapatite in ungulate mammals and implications for ecological and paleoecological studies. Oecologia 120(3), 347-363. 173 Cerling, T.E., Harris, J.M., Passey, B.H., 2003. Diets of East African Bovidae based on stable isotope analysis. Journal of Mammalogy 84(2), 456-470.

Cerling, T.E., Harris, J.M., MacFadden, B.J., Leakey, M.G., Quade, J., Eisenmann, V., Ehleringer, J.R., 1997. Global vegetation change through the Miocene/Pliocene boundary. Nature 389(6647), 153-158.

Cerling, T.E., Hay, R.L., 1986. An isotopic study of paleosol carbonates from Olduvai Gorge. Quaternary Research 25(1), 63-78.

Cerling, T.E., Quade, J., 1993. Stable carbon and oxygen isotopes in soil carbonates. In: Swart, P.K., Lohmann, J., Mckenzie, J., Savin, S., (Eds.), Climate change in Continental Isotopic Records, 217-231.

Cerling, T.E., Quade, J., Wang, Y., Bowman, J.R., 1989. Carbon isotopes in soils and palaeosols as ecology and palaeoecology indicators. Nature 341(6238), 138-139.

Cerling, T.E., Sharp, Z.D., 1996. Stable carbon and oxygen isotope analysis of fossil tooth enamel using laser ablation. Palaeogeography, Palaeoclimatology, Palaeoecology 126(1), 173- 186.

Cerling, T.E., Solomon, D.K., Quade, J., Bowman, J.R., 1991. On the isotopic composition of carbon in soil carbon dioxide. Geochimica et Cosmochimica Acta 55(11), 3403-3405.

Cerling, T.E., Wynn, J.G., Andanje, S.A., Bird, M.I., Korir, D.K., Levin, N.E., Mace, W., Macharia, A.N., Quade, J., Remien, C.H., 2011. Woody cover and hominin environments in the past 6 million years. Nature 476(7358), 51.

Churchfield, S., 1982a. A note on the diet of the African musk shrew. Acta Theriologica 27(13), 347-360.

Churchfield, S., 1982b. Food availability and the diet of the common shrew, Sorex araneus, in Britain. The Journal of Animal Ecology 15-28.

Churchfield, S., 1990. The natural history of shrews. Cornell University Press.

Churchfield, S., 2002. Why are shrews so small? The costs and benefits of small size in northern temperate Sorex species in the context of foraging habits and prey supply. Acta theriologica 47(1), 169-184.

Churchfield, S., Barrière, P., Hutterer, R., Colyn, M., 2004. First results on the feeding ecology of sympatric shrews (Insectivora: Soricidae) in the Tai National Park, Ivory Coast. Acta theriologica 49(1), 1-15.

Churchfield, S., Brown, V.K., 1987. The trophic impact of small mammals in successional grasslands. Biological Journal of the Linnean Society 31(3), 273-290. 174 Churchfield, S., Hollier, J., Brown, V.K., 1995. Population dynamics and survivorship patterns in the common shrew Sorex araneus in southern England. Acta Theriologica 40, 53-53.

Churchfield, S., Hollier, J., Brown, V.K., 1991. The effects of small mammal predators on grassland invertebrates, investigated by field exclosure experiment. Oikos 283-290.

Churchfield, S., Hollier, J., Brown, V.K., 1997a. Community structure and habitat use of small mammals in grasslands of different successional age. Journal of Zoology 242(3), 519-530.

Clark, J.D., 1993. Stone artefact assemblages from members 1–3, Swartkrans cave. Swartkrans: a cave’s chronicle of early man, pp.167-194.

Climate Research Unit, University of East Anglia. GPS 26.02 S, 27.88 E.

Codron, D., Codron, J., Lee-Thorp, J. A., Sponheimer, M., de Ruiter, D., Brink, J.S. 2007a. Stable isotope characterization of mammalian predator–prey relationships in a South African savanna. European Journal of Wildlife Research 53(3), 161-170.

Codron, D., Codron, J., Lee-Thorp, J.A., Sponheimer, M., De Ruiter, D., Sealy, J., Grant, R. Fourie, N., 2007. Diets of savanna ungulates from stable carbon isotope composition of faeces. Journal of Zoology 273(1), 21-29.

Codron, J., Codron, D., Lee-Thorp, J. A., Sponheimer, M., Kirkman, K., Duffy, K. J., Sealy, J., 2011. Landscape-scale feeding patterns of African elephant inferred from carbon isotope analysis of feces. Oecologia 165(1), 89-99.

Codron, J., Duffy, K.J., Avenant, N.L., Sponheimer, M., Leichliter, J., Paine, O., Sandberg, P., Codron, D., 2015. Stable isotope evidence for trophic niche partitioning in a South African savanna rodent community. Current Zoology 61(3): 397–41.

Codron, J., Lee-Thorp, J. A., Sponheimer, M., Codron, D., 2013. Plant stable isotope composition across habitat gradients in a semi-arid savanna: implications for environmental reconstruction. Journal of Quaternary Science 28(3), 301-310.

Codron, J., Lee-Thorp, J. A., Sponheimer, M., Codron, D., Grant, R. C., de Ruiter, D. J., 2006. Elephant (Loxodonta africana) diets in Kruger National Park, South Africa: spatial and landscape differences. Journal of Mammalogy 87(1), 27-34.

Coetzee, C.G., 1975. The biology, behaviour, and ecology of Mastomys natalensis in southern Africa. Bull World Health Organ 52, 637.

Coetzee, C.G., 1963. The prey of owls in the Kruger National Park as indicated by owl pellets collected during 1960-61. Koedoe 6(1), 115-125.

Coetzee, N., Monadjem, A., 2008. Mystromys albicaudatus. The IUCN Red List of Threatened 175 Species2008:e.T14262A4428195. http://dx.doi.org/10.2305/IUCN.UK.2008.RLTS.T14262A442 8195.en

Colvin, B.A., 1985. Barn owl foraging behavior and secondary poisoning hazard from rodenticide use on farms.

Commendador, A.S., Finney, B.P., 2016. Holocene environmental change in the eastern Snake River Plain of Idaho, USA, as inferred from stable isotope analyses of small mammals. Quaternary Research 85(3), 358-370.

Cooke, H.B.S., 1963. Pleistocene mammal faunas of Africa, with particular reference to southern Africa. African Ecology and Human Evolution 36, 65-116.

Cooke, H.B.S., 1990. Taung fossils in the University of California collections. From Apes to Angels: Essay in Anthropology in honor of Phillip V. Tobias, pp.119-134.

Coppens, Y., 1994. East side story: the origin of humankind. Scientific American 270, 88-95.

Croft, D.A., Niemi, K., Franco, A., 2011. Incisor morphology reflects diet in caviomorph rodents. Journal of Mammalogy 92, 871-879.

Curnoe, D., Grün, R., Taylor, L., Thackeray, F., 2001. Direct ESR dating of a Pliocene hominin from Swartkrans. Journal of Human Evolution 40(5), pp.379-391.

Curnoe, D., 1999. A contribution to the question of early Homo in southern Africa. Unpublished PhD Thesis, Australian National University.

Dammhahn, M., Soarimalala, V., Goodman, S. M., 2013. Trophic Niche Differentiation and Microhabitat Utilization in a Species-rich Montane Forest Small Mammal Community of Eastern Madagascar. Biotropica 45(1), 111-118.

Dart, R., 1925a. Australopithecus africanus: the man-ape of South Africa. Nature 115, 195-199.

Dart, R.A., 1925b. A note on Makapansgat: a site of early human occupation. South African Journal of Science 22, 454.

Dart, R.A., 1948. The Makapansgat proto-human australo-pithecus prometheus. American Journal of Physical Anthropology 6, 259-284.

Dart, R.A., 1949. The predatory implemental technique of Australopithecus. American Journal of Physical Anthropology 7, 1-38.

Dart, R.A., 1956. The Myth of the Bone-Accumulating Hyena. American Anthropologist 58, 40- 62. 176 Dart, R.A., 1957. The osteodontokeratic culture of Australopithecus prometheus: Transvaal Museum.

Dauphin, Y., Kowalski, C. and Denys, C., 1994. Assemblage data and bone and teeth modifications as an aid to paleoenvironmental interpretations of the open-air Pleistocene site of Tighenif (Algeria). Quaternary Research 42(3), 340-349.

Davis, D.H., 1965. Classification problems of African muridae: Balkema.

Davis, D.H.S., 1959. The barn owl's contribution to ecology and palaeoecology. Ostrich 30(S1), 144-153.

De Graaff, G., 1958. A preliminary investigation of the mammalian microfauna in Pleistocene deposits of caves in the Transvaal System.

De Graaff, G., 1961. A short survey of investigations of fossil rodents in African deposits. South African Journal of Science 57, 191-196.

De Graaff, G., 1974. Notes on the occurrence of rodents in South African National Parks. Koedoe-African Protected Area Conservation and Science 17, 173-183.

De Graaff, G., 1981. The Rodents of Southern Africa: Notes on Their Identification, Distribution, Ecology, and Taxonomy: Butterworths.

De Graaff, G., 1960. A preliminary investigation of the mammalian microfauna in Pleistocene deposits in the Transvaal System. Palaeontologia Africana 7, 59-118. de Ruiter, D.J., Sponheimer, M., Lee-Thorp, J.A., 2008. Indications of habitat association of Australopithecus robustus in the Bloubank Valley, South Africa. Journal of Human Evolution 55(6), 1015-1030. de Ruiter, P. C., Wolters, V., Moore, J. C., Winemiller, K.O., 2005. Food web ecology: playing Jenga and beyond. Science (Washington) 309(5731), 68-71.

Dean, W., 1978. Conservation of the white-tailed rat in South Africa. Biological Conservation 13, 133-140.

Delany, M., 1972. The ecology of small rodents in tropical Africa. Mammal Review 2, 1-42.

DeMenocal, P., Ortiz, J., Guilderson, T., Adkins, J., Sarnthein, M., Baker, L., Yarusinsky, M., 2000. Abrupt onset and termination of the African Humid Period: rapid climate responses to gradual insolation forcing. Quaternary Science Reviews 19, 347-361.

DeMenocal, P.B., 1995. Plio-pleistocene African climate. Science 270, 53-59. 177 DeMenocal, P.B., 2004. African climate change and faunal evolution during the Pliocene– Pleistocene. Earth Planet Science Letters 220, 3-24.

DeNiro, M., Epstein, S., 1976. You are what you eat (plus a few‰) the carbon isotope cycle in food chains. Geological Society of America 6,834.

DeNiro, M.J., Epstein S., 1978. Influence of diet on the distribution of carbon isotopes in animals. Geochim Cosmochim Acta 42, 495-506.

Denys, C., Jaeger, J., 1986. A biostratigraphic problem: the case of the East African Plio- Pleistocene rodent faunas. Modern Geology 10, 215-233.

Denys, C., 1987. Micromammals from the west Natron Pleistocene deposits (Tanzania). Biostratigraphy and paleoecology. Sciences Géologiques Bulletin 40, 185-201.

Denys, C., 1999. Of mice and men. Evolution in East and South Africa during Plio-Pleistocene times. In: Bromage, T.G., Schrenk, F., (Eds.), African Biogeography, Climate Change and Human Evolution, Oxford University Press, 226-252.

Denys, C., 1990. Implications paléoécologiques et paléobiogéographiques de l’étude de rongeurs Plio- Pleistocene d'Afrique orientale et australe. Unpubl. Thése de Doctorat d'Etat, Université Pierre et Marie Curie (Paris VI).

Denys, C., 1994. Nouvelles espèces de Dendromus (Rongeurs, Muroidea) à Langebaanweg (Pliocene, Afrique du Sud) consequences stratigraphiques et paléoécologiques. Palaeovert, 23, 153-176.

Derting, T.L., Cranford, J.A., 1989. Physical and behavioral correlates of prey vulnerability to barn owl (Tyto alba) predation. American Midland Naturalist, 11-20.

Dickman, C. R., 1988. Body size, prey size, and community structure in insectivorous mammals. Ecology 569-580.

Dickman, C. R.,1991. Mechanisms of competition among insectivorous mammals. Oecologia 85(4), 464-471.

Dodd, J.R., Stanton, R.J., 1990. Paleoecology: concepts and applications. John Wiley & Sons.Vancouver Dodd, J.R., Stanton, R.J., 1990. Paleoecology: concepts and applications. John Wiley & Sons.

Dodson, P., Wexlar, D., 1979. Taphonomic Investigations of Owl Pellets. Paleobiology 5, 275- 284.

Domínguez-Rodrigo, M., 2014. Is the “Savanna Hypothesis” a Dead Concept for Explaining the Emergence of the Earliest Hominins?. Current Anthropology 55(1), 59-81. 178 Downs, C.T., Perrin, M., 1995. The thermal biology of the white-tailed rat Mystromys albicaudatus, a cricetine relic in southern temperate African grassland. Comparative Biochemistry and Physiology Part A: Physiology 110, 65-69.

Dudu, A., Churchfield, S., Hutterer, R., 2005. Community structure and food niche relationships of coexisting rain-forest shrews in the Masako Forest, north-eastern Congo. Advances in the Biology of Shrews II New York, Special Publication International Society for Shrew Biology (1), 229-239.

Dunbar, R., Dunbar, E., 1974. Ecological relations and niche separation between sympatric terrestrial primates in Ethiopia. Folia Primatology 21, 36-60.

Ellis, R., Vogel, J. C., Fuls, A., 1980. Photosynthetic pathways and the geographical distribution of grasses in South West Africa/Namibia. South African Journal of Science 76(7), 307-314.

Evans, E.N., Van Couvering, J.A., Andrews, P., 1981. Palaeoecology of Miocene sites in western Kenya. Journal of Human Evolution 10(1), 99-116.

Faith, J.T., Behrensmeyer, A.K., 2013. Climate change and faunal turnover: testing the mechanics of the turnover-pulse hypothesis with South African fossil data. Paleobiology 39(4), 609-627.

Fernandez-Jalvo, Y., Denys, C., Andrews, P., Williams, T., Dauphin, Y., Humphrey, L., 1998. Taphonomy and palaeoecology of Olduvai Bed-I (Pleistocene, Tanzania). Journal of Human Evolution 34, 137-172.

Fernandez-Jalvo, Y., Avery, D.M., 2015. Pleistocene micromammals and their predators at Wonderwerk Cave, South Africa. African Archaeological Review 32(4), 751-791.

Fleming, T.H., 1973. Numbers of mammal species in North and Central American forest communities. Ecology 54(3), 555-563.

Fox-Dobbs, K., Bump, J.K., Peterson, R.O., Fox, D.L., Koch, P.L., 2007. Carnivore-specific stable isotope variables and variation in the foraging ecology of modern and ancient wolf populations: case studies from Isle Royale, Minnesota, and La Brea. Canadian Journal of Zoology 85(4), 458-471.

Gannes, L.Z., O'Brien, D.M., del Rio, C.M., 1997. Stable isotopes in animal ecology: assumptions, caveats, and a call for more laboratory experiments. Ecology 78(4), 1271-1276.

Gehler, A., Tütken, T., Pack, A., 2012. Oxygen and Carbon Isotope Variations in a Modern Rodent Community–Implications for Palaeoenvironmental Reconstructions. PloS One 7:e49531.

Gibbons, A., 2002. In search of the first hominids. Science 295(5558), 1214-1219. 179 Gittleman, J.L., Harvey, P.H., 1982. Carnivore home-range size, metabolic needs and ecology. Behavioral Ecology and Sociobiology 10(1), 57-63.

Glue, D.E., 1974. Food of the barn owl in Britain and Ireland. Bird Study 21(3), 200-210.

Grimes, S.T., Collinson, M.E., Hooker, J.J., Mattey, D.P., Grassineau, N.V., Lowry, D., 2004. Distinguishing the diets of coexisting fossil theridomyid and glirid rodents using carbon isotopes. Palaeogeography, Palaeoclimatology, Palaeoecology 208, 103-119.

Grimes, S.T., Collinson, M.E., Hooker, J.J., Mattey, D.P., 2008. Is small beautiful? A review of the advantages and limitations of using small mammal teeth and the direct laser fluorination analysis technique in the isotope reconstruction of past continental climate change. Palaeogeography, Palaeoclimatology, Palaeoecology 266, 39-50.

Grine, F.E., Sponheimer, M., Ungar, P.S., Lee-Thorp, J., Teaford, M.F., 2012. Dental microwear and stable isotopes inform the paleoecology of extinct hominins. American Journal of Physical Anthropology 148, 285-317.

Grine, F.E., 1993. Description and preliminary analysis of new hominid craniodental fossils from the Swartkrans Formation. Brain CK. Swartkrans: A cave’s chronicle of early man. Pretoria: Transvaal Museum, pp.75-116.

Grine, F.E., 2005. Early Homo at Swartkrans, South Africa: a review of the evidence and an evaluation of recently proposed morphs. South African Journal of Science 101(1-2), 43-52.

Hagenah, N., Prins, H.H., Olff, H., 2009. Effects of large herbivores on murid rodents in a South African savanna. Journal of Trophic Ecology 25, 483.

Hall, G., Pickering, R., Lacruz, R., Hancox, J., Berger, L. R., Schmid, P., 2006. An Acheulean handaxe from Gladysvale Cave Site, Gauteng, South Africa: research in action. South African Journal of Science 102(3 & 4), 103.

Hammer, Ø., Harper, D.A.T., 2008. Paleontological data analysis. John Wiley & Sons.

Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: Paleontological statistics software package for education and data analysis. Palaeontologia Electronica 4(1), 9pp. http://palaeo- electronica.org/2001_1/past/issue1_01.htm

Happold, D., 2013. Mammals of Africa Volume III. Rodents, Hares, and Rabbits Bloomsbury. In: Kingdon, J., Happold, D., Butynski, T., Hoffmann, M., Happold, M. and Kalina, J., (Eds.), Mammals of Africa: 6 Vols. Bloomsbury, London.

Happold, M., Happold, D.C.D., (Eds.). 2013. Mammals of Africa Volume IV. Hedgehogs, shrews and bats. In: Kingdon, J., Happold, D., Butynski, T., Hoffmann, M., Happold, M. and Kalina, J., (Eds.), Mammals of Africa: 6 Vols. Bloomsbury, London. 180 Harris, J.M., Cerling, T.E., Leakey, M.G., Passey, B.H., 2008. Stable isotope ecology of fossil hippopotamids from the Lake Turkana Basin of East Africa. Journal of Zoology 275(3), 323-331.

Hayward, M. W., Kerley, G.I., 2008. Prey preferences and dietary overlap amongst Africa's large predators. South African Journal of Wildlife Research 38(2), 93-108.

Herries, A.I., Shaw, J., 2011. Palaeomagnetic analysis of the Sterkfontein palaeocave deposits: Implications for the age of the hominin fossils and stone tool industries. Journal of Human Evolution 60(5), 523-539.

Herries, A.I., Curnoe, D., Adams, J.W., 2009. A multi-disciplinary seriation of early Homo and Paranthropus bearing palaeocaves in southern Africa. Quaternary International 202(1-2), 14-28.

Herries, A.I., Pickering, R., Adams, J.W., Curnoe, D., Warr, G., Latham, A.G., Shaw, J., 2013. A multi-disciplinary perspective on the age of Australopithecus in southern Africa. In: The paleobiology of Australopithecus. Springer, Netherlands, pp. 21-40.

Herries, A.I.R., Adams, J.W., Kuykendall, K.L., Shaw, J., 2006a. Speleology and mag- netobiostratigraphic chronology of the GD 2 locality of the Gondolin hominin- bearing paleocave deposits, NorthWest Province, South Africa. Journal of Human Evolution 51, 617- 631.

Hill, A., 1978. Taphonomical background to fossil man-problems in palaeoecology. Geological Society, London, Special Publications 6, 87-101.

Hill, A., 1979. Disarticulation and scattering of mammal skeletons. Paleobiology 261-274.

Hoffman, R., 1988. The Contribution of Raptorial Birds to Patterning in Small Mammal Assemblages. Paleobiology 14, 81-90.

Hopley, P.J., Latham, A.G., Marshall, J.D., 2006. Palaeoenvironments and palaeodiets of mid- Pliocene micromammals from Makapansgat Limeworks, South Africa: a stable isotope and dental microwear approach. Palaeogeography, Palaeoclimatology, Palaeoecology 233(3), 235- 251.

Hopley, P.J. and Maslin, M.A., 2010. Climate-averaging of terrestrial faunas: an example from the Plio-Pleistocene of South Africa. Paleobiology 36(1), 32-50.

Hughes, A.R., 1954. Hyaenas versus australopithecines as agents of bone accumulation. American Journal of Physical Anthropology 12, 467-486.

Hynek, S.A., Passey, B.H., Prado, J.L., Brown, F.H., Cerling, T.E., Quade, J., 2012. Small mammal carbon isotope ecology across the Miocene–Pliocene boundary, northwestern Argentina. Earth Planet Science Letters 321, 177-188. 181 Jaeger, J., 1976. Les rongeurs (Mammalia, Rodentia) du Pleistocene inferieur d’Olduvai Bed I (Tanzanie) 1ere Partie: Les Murides. Fossil vertebrates of Africa 4,58-120.

Jansa, S.A., Weksler, M., 2004. Phylogeny of muroid rodents: relationships within and among major lineages as determined by IRBP gene sequences. Molecular Phylogenetic Evolution 31, 256-276.

Jeffrey, A., Denys, C., Stoetzel, E. and Lee-Thorp, J.A., 2015. Influences on the stable oxygen and carbon isotopes in gerbillid rodent teeth in semi-arid and arid environments: Implications for past climate and environmental reconstruction. Earth and Planetary Science Letters 428, 84-96.

JMP®, Version Pro12. SAS Institute Inc., Cary, NC, 2015.

Keesing, F., 2000. Cryptic consumers and the ecology of an African savanna. Bioscience 50, 205-215.

Keyser, A.W., Menter, C.G., Moggi-Cecchi, J., Pickering, T.R., Berger, L.R., 2000. Drimolen: a new hominid-bearing site in Gauteng, South Africa. South African Journal of Science 96(4), 193-197.

Kimura, Y., Jacobs, L.L., Cerling, T.E., Uno, K.T., Ferguson, K.M., Flynn, L.J., Patnaik, R. 2013. Fossil Mice and Rats Show Isotopic Evidence of Niche Partitioning and Change in Dental Ecomorphology Related to Dietary Shift in Late Miocene of Pakistan. PloS One 8:e69308.

Kinahan, A.A., Pillay, N., 2008. Does differential exploitation of folivory promote coexistence in an African savanna granivorous rodent community? Journal of Mammalogy 89, 132-137.

Kingdon, J., Happold, D., Butynski, T., Hoffmann, M., Happold, M., Kalina, J., 2013. Mammals of Africa: 6 Vols. Bloomsbury, London.

Kingston, J. D., Harrison, T., 2007. Isotopic dietary reconstructions of Pliocene herbivores at Laetoli: Implications for early hominin paleoecology. Palaeogeography, Palaeoclimatology, Palaeoecology 243(3), 272-306.

Kingston, J.D., 2007. Shifting adaptive landscapes: progress and challenges in reconstructing early hominid environments. American Journal of Physical Anthropology 134(S45), 20-58.

Klein, R.G., 2009. The human career: human biological and cultural origins: University of Chicago Press.

Koch, P.L., 1998. Isotopic reconstruction of past continental environments. Annual Review of Earth and Planetary Sciences 26(1), 573-613.

Koch, P.L., Fogel, M.L., Tuross, N., 1994. Tracing the diets of fossil animals using stable isotopes. In: Michener, R., Lajtha, K. (Eds.), Stable isotopes in ecology and environmental science. John Wiley & Sons. Vancouver, 63-92. 182 Koch, P.L., Heisinger, J., Moss, C., Carlson, R.W., Fogel, M.L., Behrensmeyer, A.K., 1995. Isotopic tracking of change in diet and habitat use in African elephants. Science 267(5202), 1340-1343.

Kohn, M.J., Cerling, T.E., 2002. Stable isotope compositions of biological apatite. Reviews in mineralogy and geochemistry 48,455-488.

Kohn, M.J., McKay, M.P., Knight, J.L., 2005. Dining in the Pleistocene—Who's on the menu?. Geology 33(8), 649-652.

Kotler, B.P., Brown, J.S., Hasson, O., 1991. Factors Affecting Gerbil Foraging Behavior and Rates of Owl Predation. Ecology 72, 2249-2260.

Kotler, B.P., Brown, J.S., 1988. Environmental heterogeneity and the coexistence of desert rodents. Annual Review of Ecological Systems 281-307.

Kotler, B.P., 1984. Risk of predation and the structure of desert rodent communities. Ecology 65(3), 689-701.

Kusmer, K.D., 1990. Taphonomy of Owl Pellet Deposition. Journal of Paleontology 64, 629- 637.

Lacruz, R., Ungar, P., Hancox, P.J., Brink, J.S., Berger, L.R., 2003. Gladysvale: fossils, strata and GIS analysis. South African Journal of Science 99(5/6), 283-285.

Lavocat, R., 1956. La faune des rongeurs des grottes à australopithèques. Palaeontologica Africana 4,69-75.

Lavocat, R., 1957. Sur l'âge des faunes de Rongeurs des grottes à Australopithèques 133-134.

Lazzari, V., Charles, C., Tafforeau, P., Vianey-Liaud, M., Aguilar, J., Jaeger, J., Michaux, J., Viriot, L., 2008. Mosaic convergence of rodent dentitions. PLoS One 3:e3607.

Lecompte, E., Aplin, K., Denys, C., Catzeflis, F., Chades, M., Chevret, P., 2008. Phylogeny and biogeography of African Murinae based on mitochondrial and nuclear gene sequences, with a new tribal classification of the subfamily. BMC Evolutionary Biology 8, 199.

Lee-Thorp, J.A., Sponheimer, M., Luyt, J., 2007. Tracking changing environments using stable carbon isotopes in fossil tooth enamel: an example from the South African hominin sites. Journal of Human Evolution 53, 595-601.

Lee-Thorp, J.A., Sponheimer, M., Passey, B.H., de Ruiter, D.J., Cerling, T.E., 2010. Stable isotopes in fossil hominin tooth enamel suggest a fundamental dietary shift in the Pliocene. Philosophical Transactions of the Royal Society London B Biological Sciences 365, 3389-3396. 183 Lee-Thorp, J.A., Sponheimer, M., van der Merwe, N.J., 2003. What do stable isotopes tell us about hominid dietary and ecological niches in the Pliocene?. International Journal of Osteoarchaeology 13, 104-113.

Lee-Thorp, J., Thackeray, J.F., van der Merwe, N., 2000. The hunters and the hunted revisited. Journal of Human Evolution 39(6), 565-576.

Lee-Thorp, J.A., Sponheimer, M., Luyt, J., 2007. Tracking changing environments using stable carbon isotopes in fossil tooth enamel: an example from the South African hominin sites. Journal of Human Evolution 53(5), 595-601.

Levin, N.E., 2013. Compilation of East Africa soil carbonate stable isotope data. Integrated Earth Data Applications.

Levin, N.E., Brown, F.H., Behrensmeyer, A.K., Bobe, R., Cerling, T.E., 2011. Paleosol carbonates from the Omo Group: Isotopic records of local and regional environmental change in East Africa. Palaeogeography, Palaeoclimatology, Palaeoecology 307(1), 75-89.

Levin, N.E., Quade, J., Simpson, S.W., Semaw, S. and Rogers, M., 2004. Isotopic evidence for Plio–Pleistocene environmental change at Gona, Ethiopia. Earth and Planetary Science Letters 219(1-2), 93-110.

Levin, N.E., Simpson, S.W., Quade, J., Cerling, T.E., Frost, S.R., 2008. Herbivore enamel carbon isotopic composition and the environmental context of Ardipithecus at Gona, Ethiopia. Geological Society of America Special Papers 446, 215-234.

Louchart, A., Wesselman, H., Blumenschine, R.J., Hlusko, L.J., Njau, J.K., Black, M.T., Asnake, M. and White, T.D., 2009. Taphonomic, avian, and small-vertebrate indicators of Ardipithecus ramidus habitat. Science 326(5949), 66-66e4.

Luyt, C.J., Lee-Thorp, J.A. 2003. Carbon isotope ratios of Sterkfontein fossils indicate a marked shift to open environments c. 1.7 Myr ago: research articles: human origins research in South Africa. South African Journal of Science 99(5 & 6), 271.

MacFadden, B. J., Higgins, P., 2004. Ancient ecology of 15-million-year-old browsing mammals within C3 plant communities from Panama. Oecologia 140(1), 169-182.

Marti, C.D., 1974. Feeding Ecology of Four Sympatric Owls. The Condor 76, 45-61.

Maslin, M.A., Christensen, B., 2007. Tectonics, orbital forcing, global climate change, and human evolution in Africa: introduction to the African paleoclimate special volume. Journal of Human Evolution 53, 443-464.

Maslin, M.A., Christensen, B., 2007. Tectonics, orbital forcing, global climate change, and human evolution in Africa: introduction to the African paleoclimate special volume. Journal of Human Evolution 53(5), 443-464. 184 Maslin, M.A., Brierley, C.M., Milner, A.M., Shultz, S., Trauth, M.H., Wilson, K.E., 2014. East African climate pulses and early human evolution. Quaternary Science Reviews 101, 1-17.

Maslin, M.A., Shultz, S., Trauth, M.H., 2015. A synthesis of the theories and concepts of early human evolution. Philosophical Transactions of the Royal Society B: Biological Sciences 370(1663), 20140064.

Maslin M.A., Trauth M.H., 2009. Plio-Pleistocene East African Pulsed Climate Variability and Its Influence on Early Human Evolution. In: Grine F.E., Fleagle J.G., Leakey R.E. (Eds.), The First Humans – Origin and Early Evolution of the Genus Homo. Vertebrate Paleobiology and Paleoanthropology. Springer, Dordrecht.

Matmon, A., Ron, H., Chazan, M., Porat, N., Horwitz, L.K., 2012. Reconstructing the history of sediment deposition in caves: A case study from Wonderwerk Cave, South Africa. Bulletin 124(3-4), 611-625.

Matthews, T., Denys, C., Parkington, J.E., 2007. Community evolution of Neogene micromammals from Langebaanweg ‘E’Quarry and other west coast fossil sites, south-western Cape, South Africa. Palaeogeography, Palaeoclimatolgy, Palaeoecology 245, 332-352.

Matthews, T., Stynder DD. 2011. An analysis of the Aethomys (Murinae) community from Langebaanweg (Early Pliocene, South Africa) using geometric morphometrics. Palaeogeography, Palaeoclimatology, Palaeoecology 302, 230-242.

Mauffrey, J., Catzeflis, F., 2003. Ecological and isotopic discrimination of syntopic rodents in a neotropical rain forest of French Guiana. Journal of Tropical Ecology 19, 209-214.

McKee, J.K., 1991. Palaeo-ecology of the Sterkfontein hominids: a review and synthesis.

McLean, B.S., Emslie, S.D., 2012. Stable isotopes reflect the ecological stability of two high- elevation mammals from the late Quaternary of Colorado. Quaternary Research 77, 408-417.

Meester, J., 1963. A systematic revision of the shrew genus Crocidura in Southern Africa: The genus Crocidura. Transvaal Museum Memoirs 13(1), 8-22.

Meester, J., Meyer I.J., 1972. Fossil Suncus (Mammalia: Soricidae) from southern Africa. Annals of the Transvaal Museum 27, 269-277.

Meester, Z., 1955. Fossil shrews of South Africa. Annals of the Transvaal Museum 22, 271-278.

Mentis, M.T., 1981. Evaluation of the wheel-point and step-point methods of veld condition assessment. Proceedings of the Annual Congresses of the Grassland Society of Southern Africa 16(1), 89-94.

Miller, G.S., 1912. Catalogue of the Mammals of Western Europe:(Europe Exclusive of Russia) in the Collection of the British Museum. Order of the Trustees. 185 Miller, J.F., Millar, J.S., Longstaffe, F.J., 2008. Carbon-and nitrogen-isotope tissue-diet discrimination and turnover rates in deer mice, Peromyscus maniculatus. Canadian journal of Zoology, 86(7), 685-691.

Misonne, X., 1969. African and Indo-Australian Muridae, evolutionary trends. Annls. Mus. r. Afr. cent., Ser. 8vo., Zool., 172, pp.1-219.

Mittelbach, G.G., Steiner, C.F., Scheiner, S.M., Gross, K.L., Reynolds, H.L., Waide, R.B., Willig, M.R., Dodson, S.I., Gough, L., 2001. What is the observed relationship between species richness and productivity?. Ecology 82(9), 2381-2396.

Mondajem, A., 1997. Habitat preferences and biomasses of small mammals in Swaziland. African Journal of Ecology 35, 64-72.

Montgelard, C., Matthee, C.A., 2012. Tempo of genetic diversification in southern African rodents: The role of Plio-Pleistocene climatic oscillations as drivers for speciation. Acta Oecol 42, 50-57.

Mucina, L., Rutherford, M.C., 2006. The vegetation of South Africa, Lesotho and Swaziland. South African National Biodiversity Institute.

Murray, D.L., Boutin, S., O'Donoghue, M., 1994. Winter habitat selection by lynx and coyotes in relation to snowshoe hare abundance. Canadian Journal of Zoology 72(8), 1444-1451.

Nel, J.A.J., 1975. Species density and ecological diversity of South African mammal communities. South African Journal of Science 71(6), 168.

Nel, T.H., Henshilwood, C.S., 2016. The small mammal sequence from the c. 76–72 ka still bay levels at Blombos Cave, South Africa–Taphonomic and Palaeoecological implications for human behaviour. PloS One 11(8), p.e0159817.

Nel, T.H., Wurz, S. and Henshilwood, C.S., 2017. Small mammals from Marine Isotope Stage 5 at Klasies River, South Africa–Reconstructing the local palaeoenvironment. Quaternary International.

Passey, B.H., Cerling, T.E., 2006. In situ stable isotope analysis (δ 13 C, δ 18 O) of very small teeth using laser ablation GC/IRMS. Chemical Geology 235(3), 238-249.

Patterson, D.B., Faith, J.T., Bobe, R., Wood, B., 2014. Regional diversity patterns in African bovids, hyaenids, and felids during the past 3 million years: the role of taphonomic bias and implications for the evolution of Paranthropus. Quaternary Science Reviews 96, 9-22.

Perrin, M., Kotler, B., 2005. A test of five mechanisms of species coexistence between rodents in a southern African savanna. African Zoology 40, 55-61. 186 Pickering, R., Kramers, J.D., 2010. Re-appraisal of the stratigraphy and determination of new U- Pb dates for the Sterkfontein hominin site, South Africa. Journal of Human Evolution 59(1), 70- 86.

Pickering, R., Hancox, P.J., Lee-Thorp, J.A., Grün, R., Mortimer, G.E., McCulloch, M., Berger, L.R., 2007. Stratigraphy, U-Th chronology, and paleoenvironments at Gladysvale Cave: insights into the climatic control of South African hominin-bearing cave deposits. Journal of Human Evolution 53(5), 602-619.

Pickering, R., Kramers, J.D., Hancox, P.J., de Ruiter, D.J., Woodhead, J.D., 2011. Contemporary flowstone development links early hominin bearing cave deposits in South Africa. Earth and Planetary Science Letters 306(1-2), 23-32.

Plug, I., Keyser, A.W., 1994. A preliminary report on the bovid species from recent excavations at Gladysvale, South Africa. South African Journal of Science 90(6), 357-359.

Pocock, T., 1969. Pleistocene bird fossils from Kromdraai and Sterkfontein. Ostrich 40, 1-6.

Pocock, T., 1985. Plio-Pleistocene mammalian microfauna in southern Africa. Annals of the Geological Survey of South Africa 19, 65-67.

Pocock, T., 1987. Plio-pleistocene Fossil Mammalian Microfauna of Southern Africa: a Preliminary Report Including Description of Two New Fossil Muroid Genera (Mammalia, Rodentia). Bernard Price Institute for Palaeontological Research.

Potts, R., 1998a. Variability selection in hominid evolution. Evolutionary Anthropology: Issues, News, and Reviews 7, 81-96.

Potts, R., 1998b. Environmental hypotheses of hominin evolution. American Journal of Physical Anthropology 107, 93-136.

Potts, R., 2013. Hominin evolution in settings of strong environmental variability. Quaternary Science Reviews 73, 1-13.

Pruetz, J.D., Bertolani, P., 2009. Chimpanzee (Pan troglodytes verus) behavioral responses to stresses associated with living in a savanna-mosaic environment: implications for hominin adaptations to open habitats. PaleoAnthropology 252.

Quade, J., Cerling, T.E., Bowman, J.R., 1989. Systematic variations in the carbon and oxygen isotopic composition of pedogenic carbonate along elevation transects in the southern Great Basin, United States. Geological Society of America Bulletin 101(4), 464-475.

Radloff, F.G., Du Toit, J.T., 2004. Large predators and their prey in a southern African savanna: a predator's size determines its prey size range. Journal of Animal Ecology 73(3), 410-423. 187 Rambau, R.V., Robinson, T.J. and Stanyon, R., 2003. Molecular genetics of Rhabdomys pumilio subspecies boundaries: mtDNA phylogeography and karyotypic analysis by fluorescence in situ hybridization. Molecular Phylogenetics and Evolution 28(3), 564-575.

Rayner, R.J., Moon, B.P., Masters, J.C., 1993. The Makapansgat australopithecine environment. Journal of Human Evolution 24(3), 219-231.

Reed, D.N., Geraads, D., 2012. Evidence for a Late Pliocene faunal transition based on a new rodent assemblage from Oldowan locality Hadar AL 894, Afar Region, Ethiopia. Journal of Human Evolution 62, 328-337.

Reed, D.N., 2003. Micromammal Paleoecology: Past and present relationships between East African small mammals and their habitats. PhD thesis, Stony Brook University, Stony Brook, NY.

Reed, D.N., 2005. Taphonomic implications of roosting behavior and trophic habits in two species of African owl. Journal of Archaeological Science 32, 1669-1676.

Reed, D.N., 2011. Serengeti Micromammal Communities and the Paleoecology of Laetoli, Tanzania. In: Harrison T. (Ed.), Paleontology and Geology of Laetoli: Human Evolution in Context. Vertebrate Paleobiology and Paleoanthropology Series. Springer, Dordrechy 253-263.

Reed, K., Rector, A., 2007. African Pliocene paleoecology. In: Ungar, P.S., (Ed.) Evolution of the Human Diet. Oxford University Press on Demand, Vancouver pp. 262-288.

Reed, K.E., 1997. Early hominid evolution and ecological change through the African Plio- Pleistocene. Journal of Human Evolution 32, 289-322.

Reed, K.E., 1998. Using large mammal communities to examine ecological and taxonomic structure and predict vegetation in extant and extinct assemblages. Paleobiology 24(3), 384-408.

Reed, D.N., 2007. Serengeti micromammals and their implications for Olduvai paleoenvironments. In: Bobe, R., Alemseged, Z. and Behrensmeyer, A.K., (Eds.), Hominin environments in the East African Pliocene: an assessment of the faunal evidence. Springer Science & Business Media, Springer, Netherlands, 217-255

Reed, D.N., 2011. New Murid (Mammalia, Rodentia) Fossils from a Late Pliocene (2.4 MA) Locality, Hadar AL 894, Afar Region, Ethiopia. Journal of Vertebrate Paleontology 31(6), 1326- 1337.

Reed, K.E., 1997. Early hominid evolution and ecological change through the African Plio- Pleistocene. Journal of Human Evolution 32(2), 289-322.

Reed, K.E., Rowan, J., Kamilar, J.M., 2014. African vegetation structure: Modern analogs and hominin habitat reconstructions. American Journal of Physical Anthropology 153, 218-218. 188 Reed, K.E., Spencer, L.M., Rector, L.R., 2013. Faunal Approaches in Early Hominin Ecology. In: Sponheimer, M., Lee-Thorp, J. A., Reed, K. E., Ungar, P. (Eds.), Early Hominin Paleoecology. University Press of Colorado, pp. 3-34.

Repenning, C.A., 1967. Subfamilies and genera of the Soricidae (No. 565). US Govt. Print. Off.

Reynolds, S.C., Wilkinson, D.M., Marston, C.G., O'Regan, H.J., 2015. The 'mosaic habitat' in human evolution: past and present. Transactions of the Royal Society of South Africa, 70(1) 57- 69, http://dx.doi.org/10.1080/0035919X.2015.1007490

Robb, G.N., Woodborne, S., Bennett, N.C., 2012. Subterranean sympatry: an investigation into diet using stable isotope analysis. PLoS One, 7(11), p.e48572.

Roberts, D.L., Matthews, T., Herries, A.I., Boulter, C., Scott, L., Dondo, C., Mtembi, P., Browning, C., Smith, R.M., Haarhoff, P., Bateman, M.D., 2011. Regional and global context of the Late Cenozoic Langebaanweg (LBW) palaeontological site: West Coast of South Africa. Earth-Science Reviews 106(3-4), 191-214.

Robinson, J.T., 1954a. Prehominid dentition and hominid evolution. Evolution 324-334.

Robinson, J.T., 1954b. The genera and species of the Australopithecinae. American Journal of Physical Anthropology 12, 181-200.

Robinson, J.T., 1963. Adaptive radiation in the australopithecines and the origin of man. African ecology and human evolution 36, 385-416.

Rogers, K.L., Wang, Y., 2002. Stable isotopes in pocket gopher teeth as evidence of a Late Matuyama climate shift in the southern Rocky Mountains. Quaternary Research 57, 200-207.

Rosenzweig, M.L., 1992. Species diversity gradients: we know more and less than we thought. Journal of Mammalogy 73(4), 715-730.

Rowe-Rowe, D.T., Meester, J., 1982. Habitat preferences and abundance relations of small mammals in the Natal Drakensberg. African Zoology 17(4), 202-209.

Samuels, J.X., Van Valkenburgh, B., 2008. Skeletal indicators of locomotor adaptations in living and extinct rodents. Journal of Morphology 269, 1387-1411.

Samuels, J.X., 2009. Cranial morphology and dietary habits of rodents. Zoological Journal of the Linnean Society 156(4), 864-888.

Sare, D.T., Millar, J.S., Longstaffe, F. J., 2005. Tracing dietary protein in red-backed voles (Clethrionomys gapperi) using stable isotopes of nitrogen and carbon. Canadian Journal of Zoology 83(5), 717-725.

Scholtz, C.H., Holm, E., 1985. Insects of southern Africa. Butterworths. 189 Scott, R.S., Ungar, P.S., Bergstrom, T.S., Brown, C.A., Grine, F.E., Teaford, M.F., Walker, A., 2005. Dental microwear texture analysis shows within-species diet variability in fossil hominins. Nature 436(7051), 693.

Ségalen, L., Lee-Thorp, J.A. and Cerling, T., 2007. Timing of C4 grass expansion across sub- Saharan Africa. Journal of Human Evolution 53(5), 549-559.

Sénégas F, Avery D. 1998. New evidence for the murine origins of the Otomyinae (Mammalia, Rodentia) and the age of Bolt's Farm (South Africa). South African Journal of Science 94, 503- 507.

Sénégas F, Thackeray J. 2008. Temperature indices based on relative abundances of rodent taxa represented in South African Plio-Pleistocene assemblages. Annals of the Transvaal Museum 45, 143-144.

Sharp, Z.D., Cerling, T.E., 1996. A laser GC-IRMS technique for in situ stable isotope analyses of carbonates and phosphates. Geochimica et Cosmochimica Acta 60(15), 2909-2916.

Shipman, P. and Harris, J.M., 1988. Habitat preference and paleoecology of Australopithecus boisei in Eastern Africa. In: Evolutionary history of the robust Australopithecines. Grine, F.E. (Ed.), Routledge, pp.343-381.

Sikes, N.E., 1994. Early hominid habitat preferences in East Africa: paleosol carbon isotopic evidence. Journal of Human Evolution 27(1-3), 25-45.

Sikes, N.E., 1995. Early hominid habitat preferences in East Africa: stable isotopic evidence from paleosols (Doctoral dissertation, University of Illinois at Urbana-Champaign).

Simpson, G.G., 1960. Notes on the measurement of faunal resemblance. American Journal of Science 258(2), 300-311.

Skinner, J.D., Chimimba, C.T., 2005. The mammals of the southern African sub-region. Cambridge University Press.

Smiley, T.M., Cotton, J.M., Badgley, C., Cerling, T.E., 2016. Small-mammal isotope ecology tracks climate and vegetation gradients across western North America. Oikos 125(8), 1100-1109.

Smith, B.N., Epstein, S., 1971. Two categories of 13C/12C ratios for higher plants. Plant physiology 47(3), 380-384.

Sponheimer, M., Lee-Thorp, J.A., 1999. Isotopic evidence for the diet of an early hominid, Australopithecus africanus. Science 283, 368-370.

Sponheimer, M., Lee-Thorp, J.A., 2003. Using carbon isotope data of fossil bovid communities for palaeoenvironmental reconstruction: research articles: human origins research in South Africa. South African Journal of Science 99, 190 273-275.

Sponheimer, M., Passey, B.H., de Ruiter, D.J., Guatelli-Steinberg, D., Cerling, T.E., Lee-Thorp, J.A., 2006. Isotopic evidence for dietary variability in the early hominin Paranthropus robustus. Science 314, 980-982.

Sponheimer, M., Reed, K., Lee-Thorp, J., 2001. Isotopic palaeoecology of Makapansgat Limeworks perissodactyla. South African Journal of Science 97, 327-328.

Sponheimer, M., 1999. Isotopic paleoecology of the Makapansgat Limeworks fauna (Doctoral dissertation, Rutgers University).

Sponheimer, M., Alemseged, Z., Cerling, T.E., Grine, F.E., Kimbel, W.H., Leakey, M.G., Lee- Thorp, J.A., Manthi, F.K., Reed, K.E., Wood, B.A., Wynn, J.G., 2013. Isotopic evidence of early hominin diets. Proceedings of the National Academy of Sciences 110(26), 10513-10518.

Sponheimer, M., Lee-Thorp, J.A., 2003. Using carbon isotope data of fossil bovid communities for palaeoenvironmental reconstruction: research articles: human origins research in South Africa. South African Journal of Science 99(5 & 6), 273.

Sponheimer, M., Lee-Thorp, J.A., 2009. Biogeochemical evidence for the environments of early Homo in South Africa. In: Grine, F.E., Fleagle, J.G., Leakey, R.E. (Eds.), The first humans: origin and early evolution of the genus Homo. Springer Science & Business Media, Springer Netherlands, pp. 185-194.

Sponheimer, M., Lee-Thorp, J.A., DeRuiter, D.J., Smith, J.M., van der Merwe, N.J., Reed, K., Grant, C.C., Ayliffe, L.K., Robinson, T.F., Heidelberger, C. and Marcus, W., 2003. Diets of southern African Bovidae: stable isotope evidence. Journal of Mammalogy 84(2), 471-479.

Sponheimer, M., Robinson, T., Ayliffe, L., Passey, B., Roeder, B., Shipley, L., Lopez, E., Cerling, T., Dearing, D. Ehleringer, J., 2003. An experimental study of carbon-isotope fractionation between diet, hair, and feces of mammalian herbivores. Canadian Journal of Zoology 81(5), 871-876.

Stanford, C.B., 1996. The hunting ecology of wild chimpanzees: implications for the evolutionary ecology of Pliocene hominids. American Anthropologist 98, 96-113.

Stanley, S.M., 1992. An ecological theory for the origin of Homo. Paleobiology 18(3), 237-257.

Stratford, D., Grab, S., Pickering, T.R., 2014. The stratigraphy and formation history of fossil- and artefact-bearing sediments in the Milner Hall, Sterkfontein Cave, South Africa: New interpretations and implications for palaeoanthropology and archaeology. Journal of African Earth Sciences 96, 155-167. 191 Symes, C.T., Wilson, J.W., Woodborne, S.M., Shaikh, Z.S., Scantlebury, M., 2013. Resource partitioning of sympatric small mammals in an African forest-grassland vegetation mosaic. Austral Ecology 38(6), 721-729.

Tattersall, I., 2009. The Fossil Trail: How We Know What We Think We Know About Human Evolution. Oxford University Press.

Taylor, I., 2004. Barn owls: predator-prey relationships and conservation. Cambridge University Press.

Terry, R.C., 2010. On raptors and rodents: testing the ecological fidelity and spatiotemporal resolution of cave death assemblages. Paleobiology 36, 137-160.

Thackeray, J.F., Kirschvink, J.L., Raub, T.D., 2002. Palaeomagnetic analyses of calcified deposits from the Plio-Pleistocene hominid site of Kromdraai, South Africa: news & views. South African Journal of Science 98(11-12), 537-540.

Tieszen, L.L., Boutton, T.W., Tesdahl, K., Slade, N.A., 1983. Fractionation and turnover of stable carbon isotopes in animal tissues: implications for δ13C analysis of diet. Oecologia 57, 32-37.

Tieszen, L.L., 1991. Natural variations in the carbon isotope values of plants: implications for archaeology, ecology, and paleoecology. Journal of Archaeological Science 18, 227-248.

Tieszen, L.L., Boutton, T.W., Tesdahl, K.G., Slade, N.A., 1983. Fractionation and turnover of stable carbon isotopes in animal tissues: implications for δ13C analysis of diet. Oecologia 57(1- 2), 32-37.

Trauth, M.H., Maslin, M.A., Deino, A.L., Strecker, M.R., Bergner, A.G.N., Dühnforth, M., 2007. High- and low-latitude forcing of Plio-Pleistocene East African climate and human evolution. Journal of Human Evolution 53, 475-486.

Trauth, M.H., Deino, A.L., Bergner, A.G., Strecker, M.R., 2003. East African climate change and orbital forcing during the last 175 kyr BP. Earth and Planetary Science Letters 206(3), 297- 313.

Trauth, M.H., Larrasoaña, J.C., Mudelsee, M., 2009. Trends, rhythms and events in Plio- Pleistocene African climate. Quaternary Science Reviews 28(5), 399-411.

Trauth, M.H., Maslin, M.A., Deino, A., Strecker, M.R., 2005. Late cenozoic moisture history of East Africa. Science 309(5743), 2051-2053.

Ungar, P.S., Sponheimer, M., 2011. The diets of early hominins. Science 334, 190-193. van Couvering, J., 1980. The Encyclopedia of Paleontology. 192 Van den Heuvel, I.M., Midgley, J.J., 2014. Towards an isotope ecology of Cape Fynbos small mammals. African Zoology 49(2), 195-202. van der Merwe, N.J., Masao, F.T., Bamford, M.K., 2008. Isotopic evidence for contrasting diets of early hominins Homo habilis and Australopithecus boisei of Tanzania. South African Journal of Science 104(3-4), 153-155.

Vernon, C.J., 1972. An analysis of owl pellets collected in southern Africa. Ostrich 43(2), 109- 124.

Vézina, A.F., 1985. Empirical relationships between predator and prey size among terrestrial vertebrate predators. Oecologia 67(4), 555-565.

Vogel, J.C., 1978. Isotopic assessment of the dietary habits of ungulates. South African Journal of Science 74(8), 298-301.

Vrba, E.S., 1974. Chronological and ecological implications of the fossil Bovidae at the Sterkfontein australopithecine site. Nature 250, 19-23.

Vrba, E.S., 1975. Some evidence of chronology and palaeoecology of Sterkfontein, Swartkrans and Kromdraai from the fossil Bovidae. Nature 254, 301-304.

Vrba, E., 1974. The fossil bovidae of Sterkfontein, Swartkrans and Kromdraai (Doctoral dissertation, University of Cape Town).

Vrba ES. 1985. Ecological and adaptive changes associated with early hominid evolution. Ancestors: The Hard Evidence. Alan R.Liss, New York 63-71.

Vrba ES. 1988. 14. The Significance of Bovid Remains as Indicators of Environment and Predation Patterns. In: Behrensmeyer, A.K., (Ed.), Fossils in the making: vertebrate taphonomy and paleoecology. University of Chicago Press, pp. 247-271

Vrba, E.S., 1995. The fossil record of African antelopes (Mammalia, Bovidae) in relation to human evolution and paleoclimate. In: Vrba, E.S., Denton, G.H., Partridge, T.C., Burckle, L.H. (Eds.), Paleoclimate and evolution, with emphasis on human origins. pp. 385-424.

Waide, R.B., Willig, M.R., Steiner, C.F., Mittelbach, G., Gough, L., Dodson, S.I., Juday, G.P., Parmenter, R., 1999. The relationship between productivity and species richness. Annual review of Ecology and Systematics 30(1), 257-300.

Washburn, S.L., 1957. Australopithecines: the hunters or the hunted?. American Anthropologist 59, 612-614.

Werdelin, L., Sanders, W.J., 2010. Cenozoic mammals of Africa: University of California Press. 193 Werdelin, L., Lewis, M.E., 2005. Plio-Pleistocene Carnivora of eastern Africa: species richness and turnover patterns. Zoological Journal of the Linnean Society 144(2), 121-144.

Wesselman, H., 1995. Of mice and almost-men: regional paleoecology and human evolution in the Turkana Basin. In: Vrba, E.S., Denton, G.H., Partridge, T.C., Burckle, L.H. (Eds.), Paleoclimate and Evolution, with Emphasis on Human Origins, 356-368.

Wesselman, H.B., 1984. Omo micromammals: S. Karger.

Wesselman, H.B., 1985. Fossil micromammals as indicators of climatic change about 2.4 Myr ago in the Omo Valley, Ethiopia. South African Journal of Science 81, 260-261.

Williams, S.H., Kay, R.F., 2001. A comparative test of adaptive explanations for hypsodonty in ungulates and rodents. Journal of Mammal Evolution 8, 207-229.

Winkler, A.J., Denys, C., Avery, M.D., 2010. Rodentia. In: Werdelin, L., Sanders, W.J., (Eds.), Cenezoic Mammals of Africa. Berkeley, California: University of California Press. p 263.

Wynn, J.G., 2000. Paleosols, stable carbon isotopes, and paleoenvironmental interpretation of Kanapoi, Northern Kenya. Journal of Human Evolution 39(4), 411-432.

Wynn, J.G., 2004. Influence of Plio-Pleistocene aridification on human evolution: Evidence from paleosols of the Turkana Basin, Kenya. American Journal of Physical Anthropology 123(2), 106- 118.

Wynn, J.G., Reed, K.E., Sponheimer, M., Kimbel, W.H., Alemseged, Z., Bedaso, Z.K. and Campisano, C.J., 2016. Dietary flexibility of Australopithecus afarensis in the face of paleoecological change during the middle Pliocene: Faunal evidence from Hadar, Ethiopia. Journal of Human Evolution 99, 93-106.

Yeakel, J.D., Bennett, N.C., Koch, P.L., Dominy, N.J., 2007. The isotopic ecology of African mole rats informs hypotheses on the evolution of human diet. Proceedings of the Royal Society B: Biological Sciences 274, 1723-1730.

Yom-Tov, Y. and Wool, D., 1997. Do the contents of barn owl pellets accurately represent the proportion of prey species in the field?. Condor, pp.972-976.

Zazzo, A., Bocherens, H., Brunet, M., Beauvilain, A., Billiou, D., Mackaye, H.T., Vignaud, P. and Mariotti, A., 2000. Herbivore paleodiet and paleoenvironmental changes in Chad during the Pliocene using stable isotope ratios of tooth enamel carbonate. Paleobiology, 26(2), pp.294-309.

194 A. APPENDIX

Table A.1: Niche models for African small mammals

Land Cover and Ranks

Taxon Abbreviation Forest Woodland Bushland Grassland Semi-Arid # Taxa

Neamblysomus NEAM 0.35 0 0.25 0.4 0 2 *

Amblysomus AMBL 0.06 0.02 0.76 0.16 0 5 *

Calcochloris CALC 0.33 0.10 0.57 0.00 0.00 3 *

Chlorotalpa CHLO 0.55 0 0.1 0.35 0 2 *

Chrysochloris CHRYC 0 0 0.75 0.25 0 2 *

Chrysospalax CHRYP 1 0 0 0 0 1 * Chrysocholoridae Indet. CHRYInd. 0.38 0.02 0.41 0.19 0.00 6 * Elephantulus /Macroscelides MACR 0.00 0.13 0.22 0.16 0.49 10 *

Petrodromus PETRO 0.1 0.8 0.1 0 0 1 *

Rhynchocyon RHYN 0.80 0.10 0.07 0.00 0.00 3 *

Anomalurus ANOM 0.93 0.08 0.00 0.00 0.00 4 *

Idiurus IDIU 0.98 0.02 0.00 0.00 0.00 3 *

Zenkerella ZENK 1 0 0 0 0 1 *

Bathyergus BATH 0 0.25 0.25 0.25 0.25 2 *

Cryptomys CRYP 0 0.25 0.25 0.25 0.25 10 *

Georychus GEOR 0.25 0.25 0.25 0.25 0 1 *

Heliophobius HELI 0 0.5 0.5 0 0 1 *

Heterocephalus HETE 0 0 0 0 1 1 *

Ctenodactylus CTEN 0 0 0 0 1 *

Felovia FELO 0 0 0 0 1 *

Massoutiera MASS 0 0 0 0 1 *

Pectinator PECT 0 0 0 0 1 *

Jaculus JACU 0 0 0 0 1 *

Graphiurus GRAP 0.43 0.43 0.14 0 0 14 *

Eliomys ELIO 1 0 0 0 0 2 *

Beamys BEAM 0.5 0.5 0 0 0 1 * 195

Land Cover and Ranks

Taxon Abbreviation Forest Woodland Bushland Grassland Semi-Arid # Taxa

Cricetomys CRIC 0.625 0.375 0 0 0 2 *

Saccostomus SACC 0 0.33 0.33 0.33 0 2

Dendromus DEND 0.08 0.03 0.08 0.78 0.03

Dendroprionomys DENP 1 0 0 0 0 1 *

Malacothrix MALA 0 0 0.33 0.33 0.33 1 *

Megadendromus MEGA 1 0 0 0 0 1 *

Prionomys PRIO 1 0 0 0 0 1 *

Steatomys STEA 0 0.36 0.06 0.56 0.02 5

Mystromys MYST 0 0 0.25 0.75 0 1 *

Petromyscus PETR 0 0 0.1 0 0.9 4 *

Acomys ACOM 0.00 0.11 0.18 0.05 0.65 12

Deomys DEOM 1 0 0 0 0 1 *

Lophuromys LOPH 0.5 0 0 0.5 0 15 *

Uranomys URAN 0 0 0 1 0 1 *

Gerbillinae Indet. GERInd 0.00 0.02 0.12 0.05 0.81 9 *

Desmodilliscus DESCUS 0 0.025 0.025 0 0.95 1 *

Desmodillus DESM 0 0 0.025 0 0.975 1 *

Gerbilliscus GERBC 0.00 0.13 0.40 0.28 0.19 12

Gerbillurus GERBR 0 0 0 0 1 4 *

Gerbillus GERBS 0 0 0.2 0.2 0.6

Meriones MERI 0 0 0 0 1 3 *

Pachyuromys PACH 0 0 0 0 1 1 *

Psammomys PSAM 0 0 0 0 1 1 *

Taterillus TATE 0 0 0.43 0.00 0.57 7 *

Aethomys AETH 0.02 0.23 0.51 0.24 0.01 10

Arvicanthus ARVI 0 0 0.25 0.75 0

Colomys COLO 1 0 0 0 0 *

Dasymys DASY 0 0 0.2 0.8 0 196

Land Cover and Ranks

Taxon Abbreviation Forest Woodland Bushland Grassland Semi-Arid # Taxa

Desmomys DESMys 0 0 1 0 0 1 *

Grammomys GRAM 0.50 0.31 0.19 0.01 0.00 9

Hybomys HYBO 1 0 0 0 0 6 *

Hylomyscus HYLO 1 0 0 0 0 9 *

Lamottemys LAMO 1 0 0 0 0 1 *

Lemniscomys LEMN 0.00 0.11 0.22 0.57 0.07 9

Malacomys MALC 1 0 0 0 0 3 *

Mastomys MAST 0.00 0.14 0.33 0.53 0.00

Muriculus MURU 0 0 0 1 0 *

Mus MUS 0.18 0.34 0.11 0.35 0.02 18

Myomyscus MYOM 0.00 0.00 0.08 0.92 0.00 3 *

Oenomys OENO 0.5 0.5 0 0 0

Pelomys PELO 0 0 0.5 0.5 0

Praomys PRAO 0.8 0.2 0 0 0

Rhabdomys RHAB 0.1 0.2 0.2 0.4 0.1 1 *

Stenocephalemys STEC 0.40 0.00 0.38 0.23 0.00 4 *

Stochomys STOC 1 0 0 0 0 1 *

Thallomys THAL 0 0.5 0.5 0 0

Thamnomys THAM 1 0 0 0 0 3 *

Zelotomys ZELO 0 0.05 0.5 0.4 0.05 2 *

Otomys OTOM 0.00 0.10 0.31 0.57 0.00 13 *

Parotomys PARO 0 0 0.25 0 0.75 2 *

Petromus PETRMu 0 0 0 0 1 1 *

Thryonomys THRY 0 0 0 1 0 2 *

Tachyoryctes TACH 0 0.15 0.15 0.65 0 2 *

Soricidae Indet. SORInd 0.52 0.10 0.10 0.26 0.01 9 *

Crocidura CROC 0.21 0.21 0.21 0.32 0.07 6 *

Paracrocidura PARA 1 0 0 0 0 1 * 197

Land Cover and Ranks

Taxon Abbreviation Forest Woodland Bushland Grassland Semi-Arid # Taxa

Ruwenzorisorex RUWE 1 0 0 0 0 1 *

Scutisorex SCUT 1 0 0 0 0 1 * Suncus &/or Crocidura (small) SUN/CR 0.15 0.15 0.32 0.32 0.07 3 *

Surdisorex SURD 0 0 0 1 0 1 *

Sylvisorex SYLV 1 0 0 0 0 1 *

Congosorex CONG 0 0.33 0.33 0.33 0 1 *

Myosorex MYOS 0.33 0.25 0.05 0.37 0 5 * * new niche models

198 Table A.2: Taxonomic information and genus abbreviations. Order Family Subfamily Genus Genus Abbreviation Afrosoricida Chrysochloridae Amblysominae Neamblysomus NEAM Afrosoricida Chrysochloridae Amblysominae Amblysomus AMBL Afrosoricida Chrysochloridae Chrysochlorinae Calcochloris CALC Afrosoricida Chrysochloridae Chrysochlorinae Chlorotalpa CHLO Afrosoricida Chrysochloridae Chrysochlorinae Chrysochloris CHRYC Afrosoricida Chrysochloridae Chrysochlorinae Chrysospalax CHRYP Afrosoricida Chrysochloridae Chrysocholoridae Indet. CHRYInd. Erinacemorpha Erinacemorpha Indet ERIN Macroscelidea Macroscelididae Macroscelidinae Elephantulus/Macroscelides MACR Macroscelidea Macroscelididae Macroscelidinae Petrodromus PETRO Macroscelidea Macroscelididae Macroscelidinae Rhynchocyon RHYN Rodentia Anomaluridae Anomalurus ANOM Rodentia Anomaluridae Idiurus IDIU Rodentia Anomaluridae Zenkerella ZENK Rodentia Bathyergidae Bathyergus BATH Rodentia Bathyergidae Cryptomys CRYP Rodentia Bathyergidae Georychus GEOR Rodentia Bathyergidae Heliophobius HELI Rodentia Bathyergidae Heterocephalus HETE Rodentia Ctenodactylidae Ctenodactylus CTEN Rodentia Ctenodactylidae Felovia FELO Rodentia Ctenodactylidae Massoutiera MASS Rodentia Ctenodactylidae Pectinator PECT Rodentia Depodidae Jaculus JACU Rodentia Gliridae Graphiurus GRAP Rodentia Gliridae Eliomys ELIO Rodentia Nesomyidae Cricetomyinae Beamys BEAM Rodentia Nesomyidae Cricetomyinae Cricetomys CRIC Rodentia Nesomyidae Cricetomyinae Saccostomus SACC Rodentia Nesomyidae Dendromurinae Dendromus DEND Rodentia Nesomyidae Dendromurinae Dendroprionomys DENP Rodentia Nesomyidae Dendromurinae Malacothrix MALA Rodentia Nesomyidae Dendromurinae Megadendromus MEGA Rodentia Nesomyidae Dendromurinae Prionomys PRIO Rodentia Nesomyidae Dendromurinae Steatomys STEA Rodentia Nesomyidae Mystromyinae Mystromys MYST Rodentia Nesomyidae Mystromyinae Mystromys MYSTD Rodentia Nesomyidae Mystromyinae Proodontomys PROO Rodentia Nesomyidae Petromyscinae Stenodontomys STEN Rodentia Nesomyidae Petromyscinae Petromyscus PETR Rodentia Muridae Deomyinae Acomys ACOM Rodentia Muridae Deomyinae Deomys DEOM Rodentia Muridae Deomyinae Lophuromys LOPH Rodentia Muridae Deomyinae Uranomys URAN Rodentia Muridae Gerbillinae Gerbillinae Indet. GERInd Rodentia Muridae Gerbillinae Desmodilliscus DESCUS Rodentia Muridae Gerbillinae Desmodillus DESM Rodentia Muridae Gerbillinae Gerbilliscus GERBC Rodentia Muridae Gerbillinae Gerbillurus GERBR Rodentia Muridae Gerbillinae Gerbillus GERBS Rodentia Muridae Gerbillinae Meriones MERI Rodentia Muridae Gerbillinae Pachyuromys PACH Rodentia Muridae Gerbillinae Psammomys PSAM Rodentia Muridae Gerbillinae Taterillus TATE 199 Order Family Subfamily Genus Genus Abbreviation Rodentia Muridae Murinae Aethomys AETH Rodentia Muridae Murinae Arvicanthus ARVI Rodentia Muridae Murinae Colomys COLO Rodentia Muridae Murinae Dasymys DASY Rodentia Muridae Murinae Desmomys DESM Rodentia Muridae Murinae Golunda GOLU Rodentia Muridae Murinae Grammomys GRAM Rodentia Muridae Murinae Hybomys HYBO Rodentia Muridae Murinae Hylomyscus HYLO Rodentia Muridae Murinae Lamottemys LAMO Rodentia Muridae Murinae Lemniscomys LEMN Rodentia Muridae Murinae Malacomys MALC Rodentia Muridae Murinae Mastomys MAST Rodentia Muridae Murinae Millardia MILL Rodentia Muridae Murinae Muriculus MURU Rodentia Muridae Murinae Mus MUS Rodentia Muridae Murinae Myomyscus MYOM Rodentia Muridae Murinae Oenomys OENO Rodentia Muridae Murinae Pelomys PELO Rodentia Muridae Murinae Praomys PRAO Rodentia Muridae Murinae Rhabdomys RHAB Rodentia Muridae Murinae Saidomys SAID Rodentia Muridae Murinae Stenocephalemys STEN Rodentia Muridae Murinae Stochomys STOC Rodentia Muridae Murinae Thallomys THAL Rodentia Muridae Murinae Thamnomys THAM Rodentia Muridae Murinae Zelotomys ZELO Rodentia Muridae Myocricetodontinae Boltimys BOLT Rodentia Muridae Otomyinae Otomys OTOM Rodentia Muridae Otomyinae Euryotomys EURY Rodentia Muridae Otomyinae Parotomys PARO Rodentia Muridae Otomyinae Prototomys PROT Rodentia Petromuridae Petromus PETRMu Rodentia Thryonomyidae Thryonomys THRY Rodentia Spalcidae Tachyoryctinae Tachyoryctes TACH Soricomorpha Soricidae Soricidae Indet. SORInd Soricomorpha Soricidae Crocidurinae Crocidura CROC Soricomorpha Soricidae Crocidurinae Diplomesodon DIPL Soricomorpha Soricidae Crocidurinae Paracrocidura PARA Soricomorpha Soricidae Crocidurinae Ruwenzorisorex RUWE Soricomorpha Soricidae Crocidurinae Scutisorex SCUT Soricomorpha Soricidae Crocidurinae Suncus &/or Crocidura* SUN/CR Soricomorpha Soricidae Crocidurinae Surdisorex SURD Soricomorpha Soricidae Crocidurinae Sylvisorex SYLV Soricomorpha Soricidae Myosoricinae Congosorex CONG Soricomorpha Soricidae Myosoricinae Myosorex MYOS Soricomorpha Soricidae Myosorex &/or Crocidura** MYO/CR

200

Table A.3: Minimum Number of Individuals (MNI) for each fossil site by taxon. WW- WW FLN FLKNN HADAR OMO Genus MAK ST-M4 ST-M5 SK-M1 SK-M2 SK-M3 KB KA DR GON S12 S11 GVED ARA 1-6 1-3 SH B NEAM 0 0 7 43 29 24 0 0 0 0 0 0 0 0 0 0 0 0 AMBL 0 0 0 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 CALC 15 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CHLO 0 0 17 0 0 0 5 2 0 0 1 1 0 0 0 0 0 0 CHRYC 0 0 6 0 0 0 2 0 0 0 4 1 0 0 0 0 0 0 CHRYP 0 0 0 28 20 9 0 0 0 0 0 0 0 0 0 0 0 0 CHRYInd. 0 0 0 0 0 0 0 0 5 1 0 0 0 0 0 0 0 0 ERIN 0 0 0 0 0 0 0 0 0 0 0 0 0 4 0 0 0 0 MACR 227 7 297 511 244 266 221 10 41 6 281 222 27 0 0 0 0 0 CRYP 40 1 191 537 308 263 63 1 5 4 38 57 48 0 0 0 0 0 GEOR 0 0 6 83 25 49 38 1 0 0 0 0 0 0 0 0 0 0 HETE 0 0 0 0 0 0 0 0 0 0 0 0 0 0 21 0 0 0 GRAP 65 0 36 58 33 30 30 1 4 1 2 2 0 0 0 0 0 0 SACC 0 0 0 0 0 0 0 0 0 2 96 38 3 0 29 2 0 0 DEND 173 9 199 14 4 3 58 59 34 27 98 46 11 0 17 1 0 0 MALA 9 0 103 21 12 11 12 11 4 2 82 14 0 0 0 0 0 0 STEA 151 1 56 36 28 4 14 5 5 15 259 87 20 0 54 16 0 0 MYST 583 83 2291 3395 1133 1637 992 189 70 52 335 188 111 0 0 0 0 0 MYSTD 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PROO 155 1 6 155 52 57 63 5 13 7 0 0 1 0 0 0 0 0 STEN 246 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PETR 0 0 0 0 0 0 0 0 0 0 33 3 0 0 0 0 0 0 ACOM 295 0 2 4 2 1 13 0 18 0 0 0 7 6 0 0 250 0 URAN 0 0 0 0 0 0 0 0 0 0 0 0 0 499 0 0 0 0 GERInd 0 5 0 0 0 0 20 0 0 0 0 0 4 0 0 0 0 0 DESM 0 0 0 3 2 1 2 0 0 0 84 11 1 0 0 0 0 0 GERBC 0 1 172 76 15 33 21 3 1 2 543 286 6 54 47 15 77 7 GERBR 0 0 0 0 0 0 0 0 0 0 386 138 0 0 0 0 0 0 GERBS 0 0 0 0 0 0 0 0 0 0 0 0 0 0 65 0 0 0 TATE 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 AETH 222 1 19 98 64 68 54 2 13 11 356 153 53 0 37 23 0 0 ARVI 0 0 0 0 0 0 0 0 0 0 0 0 0 2 1 5 0 1 DASY 98 1 27 140 53 73 35 2 10 2 0 2 3 0 0 0 0 0 GOLU 0 0 0 0 0 0 0 0 0 0 0 0 0 143 0 0 270 8 GRAM 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 LEMN 0 0 0 0 0 0 11 0 0 0 6 3 1 0 0 0 0 9 MAST 60 1 79 161 78 57 6 3 4 19 82 40 22 0 23 5 0 79 MILL 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 34 0 MUS 198 1 12 24 19 6 28 18 64 31 58 15 16 8 8 3 4 1 OENO 0 0 0 0 0 0 0 0 0 0 0 0 0 24 0 6 279 0 PELO 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 3 0 0 PRAO 0 0 0 0 0 0 0 0 0 0 0 0 0 112 0 0 5 0 RHAB 173 2 59 28 18 5 11 7 8 7 24 13 4 0 0 0 0 0 SAID 0 0 0 0 0 0 0 0 0 0 0 0 0 37 0 0 327 0 THAL 0 0 0 13 2 5 1 0 0 0 2 0 0 0 4 13 0 11 201 WW- WW FLN FLKNN HADAR OMO Genus MAK ST-M4 ST-M5 SK-M1 SK-M2 SK-M3 KB KA DR GON S12 S11 GVED ARA 1-6 1-3 SH B ZELO 0 2 270 0 0 0 1 25 0 8 36 29 0 0 27 8 0 0 OTOM 829 28 621 723 288 366 250 24 77 19 174 157 68 0 89 16 0 0 THRY 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 TACH 0 0 0 0 0 0 0 0 0 0 0 0 0 49 0 0 0 0 SORInd 0 4 0 0 0 0 22 0 0 0 0 0 3 81 0 0 0 0 CROC 0 0 0 0 0 0 0 0 0 0 245 91 6 0 0 0 0 3 DIPL 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SUN/CR 95 3 57 4 1 1 1 0 15 19 28 11 11 0 0 0 0 5 MYOS 710 5 171 110 44 33 8 0 47 19 27 3 6 0 0 0 0 1 MYO/CR 0 0 0 0 0 0 0 0 0 0 225 49 0 0 0 0 0 0

202

Table A.4: Minimum Number of Individuals (MNI) for each roost site by taxon.

Genus 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 19 20 21 22 23 24 25 26 27 28

MACR 0 0 0 0 0 2 1 0 0 0 0 0 1 0 0 2 1 4 1 3 0 0 0 2 1 CRYP 0 0 0 0 0 2 0 0 19 0 19 0 0 0 0 0 1 4 1 3 0 20 0 2 1

GRAP 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 4 2 0 0 1 0 0 SACC 0 0 0 0 44 0 0 0 0 0 3 0 0 0 2 0 0 56 10 24 0 13 1 0 3 DEND 0 0 0 0 0 0 0 0 0 0 2 0 0 1 0 0 0 6 0 3 8 0 5 2 0

MALA 0 0 0 0 0 160 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 160 0 0 0 STEA 3 0 0 0 28 1 0 0 0 0 6 0 0 0 0 0 0 22 26 21 8 0 2 4 0 MYST 0 0 0 0 0 0 0 0 18 0 0 0 0 0 0 0 0 0 0 0 15 28 0 0 0 GERInd 0 0 0 0 0 0 0 0 0 0 0 44 398 64 26 0 0 0 0 0 0 0 0 0 0

DESM 0 0 0 0 0 64 62 4 40 9 0 14 2 2 14 0 0 0 0 0 0 1 0 0 0 GERBC 0 0 0 0 0 0 0 0 0 0 16 0 0 0 0 0 4 20 51 33 23 160 20 33 4 GERBS 0 0 0 0 0 77 144 0 0 43 0 0 0 0 0 28 0 0 0 0 0 0 0 0 0

AETH 0 3 0 5 11 1 1 0 0 1 3 1 1 0 0 0 8 10 9 43 1 1 2 13 0 DASY 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 3 19 MAST 94 402 54 22 42 0 4 11 19 0 1 0 0 0 0 0 3 737 200 294 130 170 0 28 1

MUS 12 10 21 0 0 1 0 2 1 4 0 0 17 0 8 0 0 55 13 18 9 91 4 0 0 RHAB 3 9 0 1 0 1 0 3 5 0 4 0 2 2 2 0 0 17 1 1 7 9 5 0 0 THAL 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0

OTOM 24 26 10 31 2 0 0 0 44 0 31 0 0 0 0 0 22 136 43 130 28 71 8 24 19 PARO 0 0 0 0 0 7 0 4 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SORInd 95 23 26 6 1 8 0 0 0 0 0 0 0 0 0 0 0 0 3 4 0 4 1 0 0 CROC 2 0 0 3 11 5 4 0 0 1 0 2 0 0 0 0 1 100 18 23 4 19 2 3 5

SUN/CR 0 0 0 0 0 3 0 0 0 0 5 0 0 0 0 0 0 17 12 13 6 15 0 0 0 MYOS 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 12 2 0 3 0

203

Table A.5: Relative abundance (%) for each fossil site by taxon. MA GO WW WW GVE FLKN FLKNN HADAR OMO Genus K ST-M4 ST-M5 SK-M1 SK-M2 SK-M3 KB KA DR N S12 S11 D ARA 1-6 1-3 SH B NEAM 0 0 0.15 0.69 1.17 0.8 0 0 0 0 0 0 0 0 0 0 0 0 AMBL 0 0 0 0 0 0 0 0 0 0 0 0 0.92 0 0 0 0 0 CALC 0.34 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CHLO 0 0 0.36 0 0 0 0.25 0.54 0 0 0.03 0.06 0 0 0 0 0 0 CHRYC 0 0 0.13 0 0 0 0.1 0 0 0 0.11 0.06 0 0 0 0 0 0 CHRYP 0 0 0 0.45 0.81 0.3 0 0 0 0 0 0 0 0 0 0 0 0 CHRYInd . 0 0 0 0 0 0 0 0 1.14 0.39 0 0 0 0 0 0 0 0 ERIN 0 0 0 0 0 0 0 0 0 0 0 0 0 0.39 0 0 0 0 11.1 MACR 5.21 4.49 6.31 8.16 9.86 8.86 5 2.72 9.36 2.36 8.02 13.37 6.19 0 0 0 0 0 CRYP 0.92 0.64 4.06 8.57 12.45 8.76 3.18 0.27 1.14 1.57 1.08 3.43 11.01 0 0 0 0 0 GEOR 0 0 0.13 1.32 1.01 1.63 1.92 0.27 0 0 0 0 0 0 0 0 0 0 HETE 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4.95 0 0 0 GRAP 1.49 0 0.77 0.93 1.33 1 1.51 0.27 0.91 0.39 0.06 0.12 0 0 0 0 0 0 SACC 0 0 0 0 0 0 0 0 0 0.79 2.74 2.29 0.69 0 6.84 1.71 0 0 16.0 10.6 DEND 3.97 5.77 4.23 0.22 0.16 0.1 2.93 3 7.76 3 2.8 2.77 2.52 0 4.01 0.85 0 0 MALA 0.21 0 2.19 0.34 0.49 0.37 0.61 2.99 0.91 0.79 2.34 0.84 0 0 0 0 0 0 STEA 3.46 0.64 1.19 0.57 1.13 0.13 0.71 1.36 1.14 5.91 7.39 5.24 4.59 0 12.74 13.68 0 0 13.3 50.0 51.3 15.9 20.4 MYST 8 53.21 48.7 54.19 45.8 54.53 5 6 8 7 9.56 11.33 25.46 0 0 0 0 0 MYSTD 0.09 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PROO 3.56 0.64 0.13 2.47 2.1 1.9 3.18 1.36 2.97 2.76 0 0 0.23 0 0 0 0 0 STEN 5.64 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PETR 0 0 0 0 0 0 0 0 0 0 0.94 0.18 0 0 0 0 0 0 ACOM 6.77 0 0.04 0.06 0.08 0.03 0.66 0 4.11 0 0 0 1.61 0.59 0 0 20.06 0 48.9 URAN 0 0 0 0 0 0 0 0 0 0 0 0 0 7 0 0 0 0 GERInd 0 3.21 0 0 0 0 1.01 0 0 0 0 0 0.92 0 0 0 0 0 DESM 0 0 0 0.05 0.08 0.03 0.1 0 0 0 2.4 0.66 0.23 0 0 0 0 0 GERBC 0 0.64 3.66 1.21 0.61 1.1 1.06 0.82 0.23 0.79 15.49 17.23 1.38 5.3 11.08 12.82 6.18 5.56 GERBR 0 0 0 0 0 0 0 0 0 0 11.01 8.31 0 0 0 0 0 0 GERBS 0 0 0 0 0 0 0 0 0 0 0 0 0 0 15.33 0 0 0 TATE 0.11 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 AETH 5.09 0.64 0.4 1.56 2.59 2.27 2.72 0.54 2.97 4.33 10.16 9.22 12.16 0 8.73 19.66 0 0 ARVI 0 0 0 0 0 0 0 0 0 0 0 0 0 0.2 0.24 4.27 0 0.79 DASY 2.25 0.64 0.57 2.23 2.14 2.43 1.77 0.54 2.28 0.79 0 0.12 0.69 0 0 0 0 0 204

MA GO WW WW GVE FLKN FLKNN HADAR OMO Genus K ST-M4 ST-M5 SK-M1 SK-M2 SK-M3 KB KA DR N S12 S11 D ARA 1-6 1-3 SH B 14.0 GOLU 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0 21.67 6.35 GRAM 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.85 0 0 LEMN 0 0 0 0 0 0 0.55 0 0 0 0.17 0.18 0.23 0 0 0 0 7.14 MAST 1.38 0.64 1.68 2.57 3.15 1.9 0.3 0.82 0.91 7.48 2.34 2.41 5.05 0 5.42 4.27 0 62.7 MILL 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2.73 0 14.6 MUS 4.54 0.64 0.26 0.38 0.77 0.2 1.41 4.89 1 12.2 1.65 0.9 3.67 0.79 1.89 2.56 0.32 0.79 OENO 0 0 0 0 0 0 0 0 0 0 0 0 0 2.36 0 5.13 22.39 0 PELO 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.47 2.56 0 0 10.9 PRAO 0 0 0 0 0 0 0 0 0 0 0 0 0 9 0 0 0.4 0 RHAB 3.97 1.27 1.25 0.45 0.73 0.17 0.55 1.9 1.83 2.76 0.68 0.78 0.92 0 0 0 0 0 SAID 0 0 0 0 0 0 0 0 0 0 0 0 0 3.63 0 0 26.24 0 THAL 0 0 0 0.21 0.08 0.17 0.05 0 0 0 0.06 0 0 0 0.94 11.11 0 8.73 ZELO 0 1.27 5.74 0 0 0 0.05 6.79 0 3.15 1.03 1.75 0 0 6.37 6.84 0 0 19.0 12.6 17.5 OTOM 2 17.83 13.2 11.54 11.64 12.19 1 6.52 8 7.48 4.96 9.46 15.6 0 20.99 13.68 0 0 THRY 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.79 TACH 0 0 0 0 0 0 0 0 0 0 0 0 0 4.81 0 0 0 0 SORInd 0 2.56 0 0 0 0 1.11 0 0 0 0 0 0.69 7.95 0 0 0 0 CROC 0 0 0 0 0 0 0 0 0 0 6.99 5.48 1.38 0 0 0 0 2.38 DIPL 0.11 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SUN/CR 2.18 1.92 1.21 0.06 0.04 0.03 0.05 0 3.42 7.48 0.8 0.66 2.52 0 0 0 0 3.97 16.2 10.7 MYOS 9 3.21 3.64 1.76 1.78 1.1 0.4 0 3 7.48 0.77 0.18 1.38 0 0 0 0 0.79 MYO/CR 0 0 0 0 0 0 0 0 0 0 6.42 2.95 0 0 0 0 0 0

205

Table A.6: Relative Abundance (%) for each roost by taxon. Genus 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 19 20 21 22 23 24 25 26 27 28 MACR 0 0 0 0 0 0.6 0.46 0 0 0 0 0 0.24 0 0 6.67 2.5 0.34 0.26 0.49 0 0 0 1.75 0 CRYP 0 0 0 0 0 0.6 0 0 13.01 0 20.88 0 0 0 0 0 2.5 0.34 0.26 0.49 0 2.62 0 1.75 0 GRAP 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.17 1.02 0.32 0 0 1.96 0 0 SACC 0 0 0 0 31.65 0 0 0 0 0 3.3 0 0 0 3.51 0 0 4.72 2.55 3.9 0 1.7 1.96 0 0 DEND 0 0 0 0 0 0 0 0 0 0 2.2 0 0 1.45 0 0 0 0.51 0 0.49 3.19 0 9.8 1.75 0 MALA 0 0 0 0 0 48.19 0 0 0 0 0 0 0 0 8.77 0 0 0 0 0 0 20.94 0 0 0 STEA 1.29 0 0 0 20.14 0.3 0 0 0 0 6.59 0 0 0 0 0 0 1.85 6.63 3.41 3.19 0 3.92 3.51 0 MYST 0 0 0 0 0 0 0 0 12.33 0 0 0 0 0 0 0 0 0 0 0 5.98 3.66 0 0 0 GERInd 0 0 0 0 0 0 0 0 0 0 0 70.97 94.54 92.75 45.61 0 0 0 0 0 0 0 0 0 0 DESM 0 0 0 0 0 19.28 28.7 16.67 27.4 15 0 22.58 0.48 2.9 24.56 0 0 0 0 0 0 0.13 0 0 0 GERBC 0 0 0 0 0 0 0 0 0 0 17.58 0 0 0 0 0 10 1.69 13.01 5.36 9.16 20.94 39.22 28.95 0 GERBS 0 0 0 0 0 23.19 66.67 0 0 71.67 0 0 0 0 0 93.33 0 0 0 0 0 0 0 0 0 AETH 0 0.63 0 7.35 7.91 0.3 0.46 0 0 1.67 3.3 1.61 0.24 0 0 0 20 0.84 2.3 6.98 0.4 0.13 3.92 11.4 0 DASY 0 0 0.89 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.16 0 0 0 0 0.89 MAST 40.34 84.99 48.21 32.35 30.22 0 1.85 45.83 13.01 0 1.1 0 0 0 0 0 7.5 62.14 51.02 47.73 51.79 22.25 0 24.56 48.21 MUS 5.15 2.11 18.75 0 0 0.3 0 8.33 0.68 6.67 0 0 4.04 0 14.04 0 0 4.64 3.32 2.92 3.59 11.91 7.84 0 18.75 RHAB 1.29 1.9 0 1.47 0 0.3 0 12.5 3.42 0 4.4 0 0.48 2.9 3.51 0 0 1.43 0.26 0.16 2.79 1.18 9.8 0 0 THAL 0 0 0 0 0 0 0 0 0 0 0 1.61 0 0 0 0 0 0 0 0 0 0 0 0 0 OTOM 10.3 5.5 8.93 45.59 1.44 0 0 0 30.14 0 34.07 0 0 0 0 0 55 11.47 10.97 21.1 11.16 9.29 15.69 21.05 8.93 PARO 0 0 0 0 0 2.11 0 16.67 0 3.33 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SORInd 40.77 4.86 23.21 8.82 0.72 2.41 0 0 0 0 0 0 0 0 0 0 0 0 0.77 0.65 0 0.52 1.96 0 23.21 CROC 0.86 0 0 4.41 7.91 1.51 1.85 0 0 1.67 0 3.23 0 0 0 0 2.5 8.43 4.59 3.73 1.59 2.49 3.92 2.63 0 SUN/CR 0 0 0 0 0 0.9 0 0 0 0 5.49 0 0 0 0 0 0 1.43 3.06 2.11 2.39 1.96 0 0 0 MYOS 0 0 0 0 0 0 0 0 0 0 1.1 0 0 0 0 0 0 0 0 0 4.78 0.26 0 2.63 0

206

Table A.7: Presence/Absence data for each biozone by taxon. Genus SAH SEL SUD GUI FOR MON EFS MAS ZBZ HIG NAM KAL CAPE NEAM 0 0 0 0 0 0 0 0 1 0 0 0 0 AMBL 0 0 0 0 0 0 0 0 0 1 0 0 1 CALC 0 0 0 0 1 0 0 0 1 0 0 0 0 CHLO 0 0 0 0 0 0 0 0 0 0 0 1 1 CHRYC 0 0 0 0 0 0 1 0 0 0 0 1 1 CHRYP 0 0 0 0 1 0 0 0 0 1 0 0 0 CHRYInd. 0 0 0 0 0 0 1 0 0 0 0 1 1 ERIN 1 0 1 1 1 0 1 1 1 0 0 0 0 MACR 0 0 0 0 1 0 1 1 1 1 1 1 1 PETRO 0 0 0 0 1 0 1 0 1 0 0 0 0 RHYN 0 0 0 0 1 1 1 1 1 0 0 0 0 ANOM 0 0 0 0 1 1 1 0 0 0 0 0 0 IDIU 0 0 0 0 1 0 0 0 0 0 0 0 0 ZENK 0 0 0 0 1 0 0 0 0 0 0 0 0 BATH 0 0 0 0 0 0 0 0 0 0 0 1 0 CRYP 0 0 0 0 0 0 0 1 1 1 0 1 1 GEOR 0 0 0 0 0 0 0 0 0 0 0 1 1 HELI 0 0 0 0 0 0 0 0 1 0 0 0 0 HETE 0 0 0 0 0 0 0 1 0 0 0 0 0 CTEN 1 0 0 0 0 0 0 0 0 0 0 0 0 FELO 0 1 1 0 0 0 0 0 0 0 0 0 0 MASS 1 0 0 0 0 0 0 0 0 0 0 0 0 PECT 0 0 0 0 0 0 0 1 0 0 0 0 0 JACU 1 1 0 0 0 0 0 0 0 0 0 0 0 GRAP 0 0 1 1 1 1 1 1 1 1 0 1 1 ELIO 1 0 0 0 0 0 0 0 0 0 0 0 0 BEAM 0 0 0 0 0 0 0 1 0 0 0 0 0 CRIC 0 0 1 1 1 1 1 1 1 0 0 0 0 SACC 0 0 0 0 0 0 1 1 1 0 1 1 1 DEND 0 0 0 0 0 1 0 1 1 1 0 1 1 DENP 0 0 0 0 1 0 0 0 0 0 0 0 0 MALA 0 0 0 0 0 0 0 0 0 1 0 1 0 MEGA 0 0 0 0 0 0 0 1 0 0 0 0 0 PRIO 0 0 0 0 1 0 0 0 0 0 0 0 0 STEA 0 0 1 1 0 0 1 0 1 1 1 1 0 MYST 0 0 0 0 0 0 0 0 0 1 0 0 1 MYSTD 0 0 0 0 0 0 0 0 0 0 0 0 0 207 Genus SAH SEL SUD GUI FOR MON EFS MAS ZBZ HIG NAM KAL CAPE PROO 0 0 0 0 0 0 0 0 0 0 0 0 0 STEN 0 0 0 0 0 0 0 0 0 0 0 0 0 PETR 0 0 0 0 0 0 0 0 0 0 1 1 0 ACOM 1 1 1 1 0 0 0 1 1 0 0 0 1 DEOM 0 0 0 0 1 1 1 0 0 0 0 0 0 LOPH 0 0 0 0 1 1 1 0 0 0 0 0 0 URAN 0 0 0 1 1 0 1 0 1 0 0 0 0 GERInd 1 1 1 1 0 0 1 1 1 1 1 1 1 DESCUS 0 1 0 0 0 0 0 0 0 0 0 0 0 DESM 0 0 0 0 0 0 0 0 0 0 1 1 0 GERBC 0 1 1 1 0 0 1 1 1 1 1 1 1 GERBR 0 0 0 0 0 0 0 0 0 0 1 1 1 GERBS 1 1 1 0 0 0 0 1 0 0 0 0 0 MERI 1 0 0 0 0 0 0 0 0 0 0 0 0 PACH 1 0 0 0 0 0 0 0 0 0 0 0 0 PSAM 1 0 0 0 0 0 0 0 0 0 0 0 0 TATE 0 1 1 1 0 0 0 1 0 0 0 0 0 AETH 0 0 0 1 0 0 1 1 1 1 1 1 1 ARVI 0 1 1 1 1 0 1 1 0 0 0 0 0 COLO 0 0 0 0 1 1 1 1 0 0 0 0 0 DASY 0 0 1 1 1 0 1 1 0 0 0 0 1 DESM 0 0 0 0 0 0 0 1 0 0 0 0 0 GOLU 0 0 0 0 0 0 0 0 0 0 0 0 0 GRAM 0 0 1 0 0 0 1 1 1 0 0 0 0 HYBO 0 0 0 0 1 1 0 0 0 0 0 0 0 HYLO 0 0 0 0 1 1 0 0 0 0 0 0 0 LAMO 0 0 0 0 0 1 0 0 0 0 0 0 0 LEMN 0 0 1 1 0 0 1 1 1 0 0 0 0 MALC 0 0 0 0 1 1 0 0 0 0 0 0 0 MAST 0 1 1 1 1 0 1 1 1 1 0 1 0 MILL 0 0 0 0 0 0 0 0 0 0 0 0 0 MURU 0 0 0 0 0 0 0 1 0 0 0 0 0 MUS 0 1 1 1 1 0 1 1 1 1 1 1 1 MYOM 0 0 0 0 0 0 0 0 0 0 0 0 1 OENO 0 0 0 0 1 1 1 1 0 0 0 0 0 PELO 0 0 0 0 0 0 1 1 1 0 0 0 0 PRAO 0 0 0 1 1 1 1 1 0 0 0 0 1 RHAB 0 0 0 0 0 0 0 1 1 1 1 1 1 SAID 0 0 0 0 0 0 0 0 0 0 0 0 0 208 Genus SAH SEL SUD GUI FOR MON EFS MAS ZBZ HIG NAM KAL CAPE STEN 0 0 1 1 0 0 1 1 0 0 0 0 0 STOC 0 0 0 0 1 0 0 0 0 0 0 0 0 THAL 0 0 0 0 0 0 0 1 1 1 1 1 0 THAM 0 0 0 0 1 1 1 0 0 0 0 0 0 ZELO 0 0 0 0 0 0 1 1 1 0 0 1 0 BOLT 0 0 0 0 0 0 0 0 0 0 0 0 0 OTOM 0 0 0 0 0 1 1 1 1 1 1 1 1 EURY 0 0 0 0 0 0 0 0 0 0 0 0 0 PARO 0 0 0 0 0 0 0 0 0 0 1 1 0 PROT 0 0 0 0 0 0 0 0 0 0 0 0 0 PETRMu 0 0 0 0 0 0 0 0 0 0 1 1 0 THRY 0 0 0 0 0 0 1 1 0 0 0 0 0 TACH 0 0 0 1 1 0 1 1 1 1 0 0 0 SORInd 1 1 1 1 1 1 1 1 1 1 1 1 1 CROC 1 1 1 1 1 1 1 1 1 1 1 1 1 DIPL 0 0 0 0 0 0 0 0 0 1 0 0 0 PARA 0 0 0 0 1 1 0 0 0 0 0 0 0 RUWE 0 0 0 0 1 1 0 0 0 0 0 0 0 SCUT 0 0 0 0 1 1 1 0 0 0 0 0 0 SUN/CR 0 0 0 0 0 1 1 1 1 1 0 0 1 SURD 0 0 0 0 0 1 0 0 0 0 0 0 0 SYLV 0 0 0 0 1 1 1 0 0 0 0 0 0 CONG 0 0 0 0 1 1 0 0 0 0 0 0 0 MYOS 0 0 0 0 0 1 1 0 1 1 0 1 1 MYO/CR 0 0 0 0 0 1 1 0 1 1 0 1 1

209

Table A.8: Presence/Absence data for each fossil site by taxon. ST- ST- SK- SK- SK- WW- WW- FLKN FLKNN HADAR OMO Genus LBW BF MAK TAUNG M4 M5 M1 M2 M3 KB KA DR GON S12 S11 GVED ARA 1-6 1-3 - SH B NEAM 0 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 AMBL 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 CALC 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CHLO 0 0 0 1 0 1 0 0 0 1 1 0 0 1 1 0 0 0 0 0 0 CHRYC 1 0 0 1 0 1 0 0 0 1 0 0 0 1 1 0 0 0 0 0 0 CHRYP 0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 CHRYInd. 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 ERIN 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 MACR 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 BATH 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CRYP 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 GEOR 0 0 0 1 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 HETE 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 GRAP 1 0 1 1 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 SACC 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 0 1 1 0 0 DEND 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 MALA 0 0 1 1 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 STEA 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 MYST 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 MYSTD 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PROO 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 0 0 0 0 STEN 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PETR 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 ACOM 1 1 1 1 0 1 1 1 1 1 0 1 0 0 0 1 1 0 0 1 0 URAN 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 GERInd 0 1 0 0 1 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 DESM 1 0 0 0 0 0 1 1 1 1 0 0 0 1 1 1 0 0 0 0 0 GERBC 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 GERBR 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 GERBS 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 TATE 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 AETH 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 ARVI 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 1 DASY 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 0 0 0 GOLU 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 1 GRAM 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 LEMN 0 0 0 0 0 0 0 0 0 1 0 0 0 1 1 1 0 0 0 0 1 MAST 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 MILL 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 MUS 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 OENO 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 0 PELO 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 PRAO 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 RHAB 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 SAID 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 210 ST- ST- SK- SK- SK- WW- WW- FLKN FLKNN HADAR OMO Genus LBW BF MAK TAUNG M4 M5 M1 M2 M3 KB KA DR GON S12 S11 GVED ARA 1-6 1-3 - SH B THAL 1 0 0 0 0 0 1 1 1 1 0 0 0 1 0 0 0 1 1 0 1 ZELO 1 0 0 1 1 1 0 0 0 1 1 0 1 1 1 0 0 1 1 0 0 BOLT 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 OTOM 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 THRY 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 TACH 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 SORInd 1 0 0 0 1 0 0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 CROC 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 1 DIPL 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SUN/CR 1 0 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 0 0 0 1 MYOS 1 0 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 0 0 0 1 MYO/CR 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0

211

Table A.9: Presence/Absence data for each roost site by taxon. Genus 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 19 20 21 22 23 24 25 26 27 28 MACR 0 0 0 0 0 1 1 0 0 0 0 0 1 0 0 1 1 1 1 1 0 0 0 1 1 CRYP 0 0 0 0 0 1 0 0 1 0 1 0 0 0 0 0 1 1 1 1 0 1 0 1 1 GRAP 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 1 0 0 SACC 0 0 0 0 1 0 0 0 0 0 1 0 0 0 1 0 0 1 1 1 0 1 1 0 1 DEND 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 1 0 1 1 0 1 1 0 MALA 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 STEA 1 0 0 0 1 1 0 0 0 0 1 0 0 0 0 0 0 1 1 1 1 0 1 1 0 MYST 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 GERInd 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 DESM 0 0 0 0 0 1 1 1 1 1 0 1 1 1 1 0 0 0 0 0 0 1 0 0 0 GERBC 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 1 1 1 1 1 1 1 1 GERBS 0 0 0 0 0 1 1 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 AETH 0 1 0 1 1 1 1 0 0 1 1 1 1 0 0 0 1 1 1 1 1 1 1 1 0 DASY 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 MAST 1 1 1 1 1 0 1 1 1 0 1 0 0 0 0 0 1 1 1 1 1 1 0 1 1 MUS 1 1 1 0 0 1 0 1 1 1 0 0 1 0 1 0 0 1 1 1 1 1 1 0 0 RHAB 1 1 0 1 0 1 0 1 1 0 1 0 1 1 1 0 0 1 1 1 1 1 1 0 0 THAL 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 OTOM 1 1 1 1 1 0 0 0 1 0 1 0 0 0 0 0 1 1 1 1 1 1 1 1 1 PARO 0 0 0 0 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SORInd 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 1 0 0 CROC 1 0 0 1 1 1 1 0 0 1 0 1 0 0 0 0 1 1 1 1 1 1 1 1 1 SUN/CR 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 1 1 1 1 1 0 0 0 MYOS 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 1 0 1 0 212

Figure A.1: Individual rarefaction curves for southern and eastern fossil sites included in this study with estimated species richness and 95% confidence interval.

Figure A.2: Individual rarefaction curves for eastern fossil sites only with estimated species richness and 95% confidence interval.