African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants

Quaternary International xxx (2017) 1e23

Contents lists available at ScienceDirect

Quaternary International journal homepage: www.elsevier.com/locate/quaint

African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants Carlos Cordova a, *, Graham Avery b, c a b c

Department of Geography, Oklahoma State University, Stillwater, OK, USA Iziko South African Museum, Natural History Collections Department: Cenozoic Studies Section, Cape Town, South Africa University of Cape Town, Archaeology Department, Cape Town, South Africa

a r t i c l e i n f o

a b s t r a c t

Article history: Received 8 June 2016 Received in revised form 22 December 2016 Accepted 26 December 2016 Available online xxx

This study tests the association between opal phytoliths in dental calculus on modern, historic, and prehistoric specimens of Loxodonta africana (African savanna elephant) with their local and regional vegetation. The modern samples were obtained from dental remains from deceased animals at the Addo Elephant National Park (Eastern Cape Province) and the Pilanesberg National Park & Game Reserve (Northwest Province) in the Republic of South Africa. The historic and prehistoric specimens, presumed to be free-roaming elephants, were sampled from museum collections in the Eastern Cape and Western Cape Provinces. In addition to comparing phytolith assemblages in dental calculus with those of the main vegetation associations, this study assesses the phytolith assemblage differences between free-roaming and park elephants. The results show that: (1) the phytolith assemblages in dental calculus of park elephants show little variation among individual specimens and close resemblance to phytolith assemblages of soils inside their areas of confinement; (2) the free-roaming specimens have a much higher diversity of phytolith morphotypes than those in parks and reserves, exhibiting sometimes typical signatures of more than one biome; (3) free-roaming Cape elephants from fynbos areas have significant amounts of Restionaceae phytoliths, which suggests that grazing on restios in grass-poor fynbos types was important; (4) short saddles, typical of Chloridoideae grasses, are always the most abundant short-cell morphotypes in dental samples, even in areas where other grass subfamilies dominate, and (5) with some limitations, the study of phytoliths in herbivore dental calculus has a high, largely unexplored, potential in paleoecology and conservation ecology. © 2017 Elsevier Ltd and INQUA. All rights reserved.

Keywords: Phytoliths African elephant South Africa Cape Region Biomes Paleoecology

1. Introduction Opal phytoliths are microscopic particles of amorphous silica deposited in cellular and extra-cellular parts of plants by means of absorbed silica in soluble state from underground water (Piperno, 2006). Eventually, silica becomes solid and resistant to organic decay, being sometimes the only part of a plant surviving as fossils. Most of these particles take the form of characteristic cells and other parts of the plant, rendering these forms useful to identify particular taxonomic groups of plants. The opal phytolith approach

* Corresponding author. E-mail addresses: [email protected] (C. Cordova), [email protected] (G. Avery).

has strong applications in archaeology, and their usefulness in paleoecology has provided information on long-term vegetation structure (e.g. Alexandre et al., 1997; Blinnikov, 2005; Golyeva, 2007; Bremond et al., 2008; Neumann et al., 2009; Cordova et al., 2011), paleoclimates (e.g., Fredlund and Tieszen, 1994; Lu et al., 2006, 2007), and definition of stages of plant-herbivore coevolu€mberg, 2004; Prasad et al., tion in deep geologic time (e.g., Stro 2005). In general, opal phytoliths can provide paleo-vegetation information including taxonomic details of C3 and C4 grasses (e.g., Twiss et al., 1969; Fredlund and Tieszen, 1994), graminoids, and a number of monocots and dicots (Runge, 1999; Piperno, 2006; Neumann et al., 2009; Mercader et al., 2010), as well gymnosperms and other groups of higher plants (e.g., Klein and Geis, 1978; Kondo and

http://dx.doi.org/10.1016/j.quaint.2016.12.042 1040-6182/© 2017 Elsevier Ltd and INQUA. All rights reserved.

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

2

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Tsumida, 1978; Blinnikov, 2005). Opal phytoliths provide a potential alternative proxy for paleovegetation reconstruction particularly in deposits where pollen grains are absent or do not preserve (Scott, 2002). However, it is important to bear in mind that phytolith records do not replace the type of information provided by pollen analysis, since they both represent different aspects of vegetation. The research approach that uses opal phytoliths embedded in dental calculus (i.e., tartar) can provide relevant information on herbivore diets and their paleoenvironments. Unfortunately, this approach is still in its infancy, limited mostly to tests that emphasize its potential for paleodiet reconstruction (e.g., Armitage, 1975; Middleton and Rovner, 1994; Gobetz and Bozarth, 2001), particularly on specimens of extinct North American Pleistocene megaherbivores (Bozarth and Hofman, 1998; Gobetz and Bozarth, 2001; Scott-Cummings and Albert, 2007; Cordova and Agenbroad, 2009). Nonetheless, the potential for its application to prehistoric and modern herbivores can be tested in modern and historical environments where fauna-vegetation relations are known, and studies of modern phytolith assemblages exist. Thus, the study presented here is an attempt to relate modern (reserve) and free-roaming (prehistoric and historic) African elephants with vegetation associations and biomes using phytoliths assemblages from soils and from dental calculus of a number of specimens. The Republic of South Africa contains nine different biomes (Fig. 1a), and historical and archaeological evidence suggest that elephants once roamed in all of them (Ebedes et al., 1995), although not all of them are considered preferred elephant habitat (Boshoff and Kerley, 2001). At present, however, elephants in the Republic of South Africa are confined to parks and reserves (Fig. 1b). The study presented here focuses mainly on the southern part of the country, where reports of elephants and other megafauna go back to the mid-1600s, when the Dutch established a colony near the Cape of Good Hope and from where European settlement radiated into the interior and along the coast (Fig. 2). Thus Dutch settlers and explorers in the 17th and 18th century and later British settlers, missionaries, and explorers of the 19th century produced historical records all of which can be used to describe herds of elephants being exterminated (Skead et al., 2007; Boshoff and Kerley, 2010). Based on historical records, elephants in the Greater Cape region existed in a mostly non-savanna environment, including various vegetation types of fynbos, renosterveld, coastal succulent scrub, forest, subtropical thicket, grassland, nama karoo and succulent karoo (Fig. 1a), although not all of these environments sustained elephants permanently (Boshoff and Kerley, 2001). The fact that the so-called “savanna elephant” (i.e., Loxodonta africana var. africana) adapted to non-savanna vegetation communities is an interesting topic discussed by several authors (Carter, 1970; Ebedes et al., 1995; Seydack et al., 2000; Boshoff and Kerley, 2001; Skead et al., 2007) and part of the present research. Furthermore, the fact that some habitats in the Cape region provided permanent elephant habitats whereas others were only seasonal (per Boshoff and Kerley, 2001), meant that specific migration patterns would have existed. Some migratory routes could be hypothesized using proxies such as stable isotopes or in the case of this study through distinctive phytolith assemblages associated with particular flora. Within this contextual framework, the study presented here is a preliminary research project with the following objectives: (1) assess dietary differences between freeroaming and park elephants, (2) assess the diet of the savanna elephant in non-savanna biomes of temperate southern Africa, and (3) to assess the use of the opal phytoliths in dental calculus to study prehistoric elephant-vegetation interactions. Objective 1 is applied to the Addo Elephant National Park in the Eastern Cape Province coastal region and the Pilanesberg National Park & Game

Reserve in the Northwest Province. Objectives 2 and 3 directs attention to phytoliths in soil and dental calculus samples from a broad area in the Cape Region (Focus area in Fig. 1a and b). 2. Background information 2.1. Study areas The focus area of this study comprises the southern part of the Western Cape and Eastern Cape Provinces south of latitude 32
S (Fig. 1b). In general terms, this area constitutes the southern part of the Greater Cape Floristic Region. Its physiography is dominated by landforms resulting from the tectonic and erosional evolution of Cape Fold structures and the African Escarpment (Partridge, 1998; Maud, 2012). The Cape Fold structures form a series of mountain systems parallel to the coast, which have created a series of interior valleys connected by antecedent stream valleys. Some of the mountain systems associated with the Cape Fold Belt include the Cederberg, the Hottentots-Holland, the Outeniqua Mountains and the Suurberg. The African Escarpment forms the edge of a plateau in the west forming a series of mountains such as the Roggeveldberg, the Nuweveldberg and the Sneeuberg (Fig. 2). Towards the east, the Escarpment is marked by the Drakensberg Mountains. Other erosional remnants of the retreating African Escarpment have formed minor mountain systems such as the Amathole Mountains in the Eastern Cape. Between the Cape Fold Mountains and the African Escarpment lies a vast area of interior plains and lowlands drained by the Gamka, Groote and Sundays Rivers. The lithology of the Cape Fold Mountains includes sandstone and shale, with minor structures of granites, quartzite, conglomerates, and limestone. The Escarpment consists of even older rocks, mostly sedimentary, with a few of volcanic origin. The coastal areas in the west and south are dominated by plains and rolling hills of eolian sand, eolianites and alluvium of Late Cenozoic age. The focus area of this study encompasses the southern part of South Africa's winter rainfall zone (WRZ), an area defined by more than the 60% in the winter months, the summer rainfall zone, with less than 40% of winter rain, (SRZ), and the all-year rainfall zone (ARZ), with between 40 and 60% of rain in winter (Fig. 1). The latter includes an area with rains uniformly distributed throughout the year on the south coast, and an area with two conspicuous rainfall concentrations (i.e., bimodal) in winter and summer rainfall. The € ppen climate types in the focus area include the Cs (MediterraKo nean) and Cf (temperate with rain all year) in the west and coastal south, BS (semiarid) in the interior, and Cw (temperate with summer rains) in the east. Temperatures in the focus area are highly modified by elevation, solar irradiation (dependent on cloud cover), ocean temperatures, and locally, on slope orientation and exposure. Areas of the south coast with frequent cloudy days tend to have less extreme temperatures, while those in the dry interior exhibit a wide range daily and seasonally. Winter snow is common at high elevations, usually above 1000 m in the west and 1800 m in the east. Territories of seven of the nine biomes present in South Africa are included in the focus area (Fig. 1). The fynbos biome, which dominates the WRZ and parts of the ARZ, consists mainly of shrubby vegetation, which in some cases can be dominated by Proteaceae (proteoid fynbos), Ericaceae (ericoid fynbos), or Restionaceae (restioid fynbos) (Campbell, 1986). Grasses are rare in most fynbos types, except in types in the ARZ, where summer rainfall has higher incidence, particularly on quartzite or shale substrate (Rebelo et al., 2006). Most fynbos associations thrive on nutrientpoor soils developed on sand, sandstone, granite, and limestone. In contrast, renosterveld (a type of fynbos association) develops on

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Fig. 1. Biomes of South Africa, indicating focus area and two studied parks (b), and parks and reserves with elephant populations in South Africa.

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

3

4

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Fig. 2. Expansion of European settlement of the Cape Colony until the early 1800s indicating main rivers, mountain ranges and neighboring indigenous groups. Compiled by the author using dates by Guelke (1989).

nutrient-rich soils, like those developed on shale and occasionally old fine-grained alluvium. Unlike the high diversity of species in the other fynbos types, renosterveld is dominated by renosterbos (Elytroppapus rhinocerotis), a bush of the Asteraceae family. Grasses can be found in relatively high proportions in the grassy fynbos types in the ARZ and in the renosterveld. Although C3 and C4 grasses can grow in fynbos vegetation, C3 tends to be dominant, particularly in areas with the highest percent of winter rainfall. The succulent karoo occurs in the drier parts of the WRZ and areas of the ARZ with relatively high winter rainfall. This biome is characterized by succulent shrubs and herbs of the Asteraceae, Mesembryanthemaceae, Aizoaceae, and Asphodeloideae families, and other drought resistant plants. Together with fynbos, the succulent karoo has one of the highest plant diversities and endemisms on the African continent (Van Wyck and Smith, 2001). Grasses are rare in most succulent karoo types, and both C3 and C4 grasses occur depending on the proportions of winter and summer rain. The nama karoo is also dominated by drought tolerant shrubs and scrub, particularly those of the Asteraceae family such as the genera Pentzia and Felicia, and several succulents of the Crassulaceae and Aizoaceae family. In some parts, shrubs are widespread, particularly Acacia karoo, Schotia afra, and Euclea undulata. Grasses may dominate in areas with reduced grazing, particularly during rainy years. Because this biome is in the SRZ, most grasses are C4, of which the majority are in the drought-resistant Aristidodieae and Chlroidoideae subfamilies. The Albany Thicket, often referred to as subtropical thicket, is dominated by a dense cover of thorny shrubs and succulents (e.g. spekboom, Portulacaria afra). Other succulents such as euphorbias and aloes dominate in some thicket types. Interestingly, the thicket biome incorporates elements from the surrounding biomes, which is why it is sometimes considered a transitional type of vegetation and not a specific biome (Vlok et al., 2003; Hoare et al., 2006). The subtropical thicket in general has few grasses due to the high competition posed by dominant shrubs, although they are often found in areas opened up by big herbivores (e.g., elephant, rhinoceros, and kudu) or by fire (Lechmere-Oertel et al., 2005; Hoare et al., 2006). Sometimes intense grazing and fire pressure produces more open vegetation, although with low species richness (Kerley et al., 1995; Kerley and Landman, 2006). C3 and C4 grasses coexist in most subtropical thicket types, but the C4 grasses normally dominate. Subtropical thicket is the typical vegetation of the Addo Elephant National Park. The Afrotemperate forest occupies a small area of the coast and

coastal mountains in the most humid part of the ARZ. Trees such as Podocarpus latifolius, Afrocarpus falcatus, Ocotea bullata, Olea capensis subsp. macrocarpa, Pterocelastrus tricuspidatus, and Rapanea melanophloeos tend to be common throughout the forest. Grasses are rare and the existing species are mainly C3; nevertheless, given the high incidence of summer rain C4 grasses can be common, particularly in open areas. Fynbos usually occurs in clearings in the forest. Grasslands and savannas occupy a relatively small proportion of the focus area. The grassland biome is confined mainly to high elevations in the mountains in the east of the focus region. C4 grasses dominate, but C3 species are not uncommon, particularly at higher elevations. The savanna biome comprises only one type in the eastern part of the region in the former Ciskei and the Albany district around East London where communities of predominantly C4 grasses intersperse with shrubs, particularly Acacia natalitia. 2.2. The elephant reserves of this study The Addo Elephant National Park (AENP), located in the focus area of this research (Fig. 2), occupies the hills and terraces of the Sundays River, the Zuurberg Mountains, and the dune cordon along the coast. Total annual precipitation at the park is 450 mm distributed throughout the year with October, November, December, March and April as the five wettest month accumulating about half of the total for the year. Mean daily temperatures fluctuate between 28 and 17
C in February and 26- 6
C in July. Although subtropical thicket vegetation occupies most of the park, the Addo Elephant National Park includes Afromontane forest on some protected slopes of the Zuurberg Mountains and Alexandria sector of the park with grassy fynbos on the summits of the Zuurberg Mountains. As indicated by the Dutch name (Zuur, or “sour”), most of the grasses in this vegetation type correspond to the sour grasses, namely the Panicoideae subfamily. Although C4 grasses predominate (e.g., Themeda triandra, Trystachya leucothrix, and Eragrostis curvula), C3 grasses (e.g., Festuca costata, Merxmuellera stricta) are not uncommon. Because grasses co-dominate with typical fynbos plant families such as Proteaceae, Ericaceae, and Asteraceae, this type of fynbos (Zuurberg quartzite and shale fynbos) is often referred to as grassy fynbos (Rebelo et al., 2006; Esler et al., 2014). Although the elephant fenced area has expanded, at the time of this research (2008) it covered 103 km2 (out of the 1640 km2 of the entire park). The fenced area covers only subtropical thicket, classified as Sundays Subtropical Thicket and the Coega Bontveld

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

(Hoare et al., 2006). The former, which occupies most of the area of the elephants' distribution, comprises dense thickets mainly on nutrient-poor soils. The Coega Bontveld has less dense islands of thicket and more open secondary grasslands, mainly on richer soils, and combinations of elements from succulent karroo, fynbos, and grasslands (Hoare et al., 2006). The dominant species in both types are succulents such as Portulacaria afra, Aloe ferox, A. africana and a number of thorns such as Carissa bispinosa and Acacia natalitia, and small trees and tall shrubs such as Schotia afra var. afra, Euclea undulata, and Olea capensis, among others. Grasses in both vegetation types include C4 (e.g., Aristida diffusa, Cynodon dactylon, Eragrostis curvula, Panicum maximum, Heteropogon contortus, and Themeda triandra) and C3 grasses (e.g., Ehrharta calycina, Merxmuellera disticha, Helichtotricon hirtulum, H. turgidulum, and Stipa dregeana). However, the C4 grasses are the most abundant throughout the park. The Pilanesberg National Park & Game Reserve (PNPGR) is located in the Northwest Province in the region traditionally known as the western Transvaal (Fig. 2). The reserve is located in an ancient volcanic massif of intrusive and extrusive alkaline, silica-poor, and potassium-sodium-rich rocks. Elevations range between 1100 and 1500 m, and mean latitude is S 25
140 4000 . The climate is subtropical, with an annual precipitation of 600e700 mm, most of which falls between October and May. The average daily maximum and minimum temperatures at the park fluctuate between 32
C and a18
C in January and 20
C and 2
C in July. The vegetation in the reserve is broadly classified as bushveld savanna, and more specifically as the Pilanesberg Mountain Bushveld (Mucina and Rutherford, 2006). The arboreal vegetation is dominated by broadleaf trees with Combretum molle, C. apiculatum and C. zeyheri, and Croton gratissimus, among others, and various tall shrubs among which several Grewia species are important (Mucina and Rutherford, 2006). The herbaceous vegetation shows the typical bushveld dominance of C4 grasses, of which the Panicoideae are the most abundant ehence the name Sour Bushveld given by Acocks (1988). 2.3. The historical range of elephants in the Cape Region Unlike most other countries in Sub-Saharan Africa, the Republic of South Africa has no herds of free-roaming elephants; all of them are confined to parks and reserves (Blanc et al., 2007; Carruthers et al., 2008) (Fig. 1b). The Kruger National Park, Tembe Elephant Park, the Addo Elephant National Park, and the Knysna Forest (part of the Garden Route National Park) are the only parks that harbor local descendants of free-roaming elephants (Hall-Martin, 1992). Other parks and reserves have elephants descended from individuals translocated from the Kruger National Park (Garaï et al., 2004; Carruthers et al., 2008). Historical records of elephants in the Cape Region date back to the Portuguese seafarers in the late 15th and early 16th century, as well as accounts of shipwreck survivors in subsequent centuries (Carruthers et al., 2008). After the mid-17th century and the Dutch settlement historical records of fauna begin to accumulate, though still scarce and in some cases elusive (Skead et al., 2011). More reliable records, however, are from the 18th century, mainly from explorers, travelers, traders, and missionaries (Skead et al., 2007, 2011; Boshoff and Kerley, 2010). In the early 19th century the extent of the Cape Colony, then under British control, encroached the largest remaining herds into areas east of the Sundays River. As the White colonization advanced from its focal point in Cape Town in the 18th century (Fig. 2), a wave of extermination advanced ahead of it. The killing of elephants for ivory, meat, conflict with farmers, and sport, decimated the populations and forced the survivors into isolated areas of dense vegetation of the Knysna Forest

5

and the Addo Bush, neither of which has soils suitable for crops. Thus, by the beginning of the 20th century, these were the only remaining areas with elephants. Despite the killings, elephants ventured out of their hideouts in forest or thicket to raid crops, resulting in even heavier responses from local human populations, sometimes through systematic shootings, as with the mass shooting of elephants by Major P.J. Pretorius in the years 1919e1920, allegedly claiming 119 elephants in the Addo Bush area alone, and 5 in the Knysna Fores (Greig, 1982; Hoffman, 1993). Two of the specimens sampled for the present study were elephant remains were the result of such killings. In the end, the elephant shootings failed to stop the raids on crops while the public became increasingly angered at the mass killings. As attempt to alleviate these problems was the creation of a park to protect the Addo Bush elephants. Thus, the precursor of the modern Addo Elephant National Park was established in 1931 with a population of only 11 elephants (Carruthers et al., 2008). Since then the population increased to 391 in 2006 necessitating expansion of the park (Carruthers et al., 2008). The elephant survivors in the Knysna Forest had a different fate. Although the dense cover of Afrotemperate forest and the rugged topography protected them, they still fell prey to humans. The main problem for the elephants in the forest was that they were living in an environment in which their diets and social relations were severely negatively affected (Seydack et al., 2000; Carruthers et al., 2008). Inevitably those elephants that were not shot began to die of malnutrition. Elephant populations in the forest proved difficult to track, but estimates indicate numbers from 20 animals in 1908 to 10 in 1970 (Carter, 1970; Milewski, 2002b). By 1990 the population had fallen to four individuals and by 2001 to three (Hall-Martin, 1992; Carruthers et al., 2008). Attempts to reintroduce populations of elephants in the 1990s failed and the native population of the forest continued to decline (Milewski, 2002b). At the time of writing only one female has been spotted in the forest (Lizette Moolman, elephant ecologist at SANparks, personal communication). The geographic range of the elephant in the southern Cape Region focus area of this study is known by historical records, archaeological and paleontological finds, and indirect evidence such as rubbing rocks and rock art compiled in maps by J.C. Skead (Skead et al., 2007, 2011) and Carl Vernon (Ebedes et al., 1995) (Fig. 3). Despite some differences these two maps of show a large concentration of elephant localities on the eastern part of the study area, including the AENP (Fig. 3 and b). In this high-concentration area, Boshoff and Kerley (2001) found a strong association of elephant localities with the presence of subtropical thicket, where spekboom (Portulacaria afra) is purportedly one of the key foods. Elsewhere, localities with prehistoric and historic elephant evidence are concentrated along the floodplains of major rivers and their tributaries, most notably the Gamtoos and the Gouritz rivers, and in some coastal areas (Fig. 3). Outside the areas described above the roaming of elephants was seasonal or occasional (Boshoff and Kerley, 2001). One can infer that the optimal habitats existed in the east of the study area and along floodplains, with marginal habitats practically everywhere else. 3. The research framework 3.1. Elephants, vegetation, and opal phytoliths The methodological framework devised for this study is based on a hypothetical model that hypothesizes the relationship between elephants and their surrounding vegetation using opal phytoliths trapped in dental calculus (Fig. 4). This should reflect the

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

6

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Fig. 3. Localities with prehistoric and historic evidence of elephants after (a) J.C. Skead (Skead et al., 2007, 2011) and (b) Carl Vernon (Ebedes et al., 1995).

diversity of phytolith morphotypes and the presence of phytoliths unique to a particular vegetation type or biome within the elephant range or inside a park. The model also suggests that park and reserve elephants should have a more limited range of opal phytoliths determined by the vegetation types within the boundaries of their confined area (Fig. 4). In the present study, testing of this model was done in two ways: first comparing dental calculus phytoliths with soil surface phytoliths in two parks: Pilanesberg National Park & Game Reserve and Addo Elephant National Park (Fig. 1b), and secondly, by comparing phytolith assemblages in modern soil surfaces in various vegetation types in the focus region (Fig. 5a) with the phytolith assemblages in the dental calculus of free-roaming elephants in the same region (Fig. 5b). 3.2. Dental calculus samples The dental calculus from the Pilanesberg National Park & Game Reserve and the Addo Elephant National Park includes samples from molars of dead elephants housed in the parks; in some, characteristics as to age and gender were recorded (Table 1). Although they constitute a small portion of the samples from each park, they elucidate the vegetation choices that these herbivores made in relation to what was available in the areas in which they were confined. It is known that elephants replace their molars every six or seven years, having gone through six or seven sets of molars at the end of adulthood. Although we acknowledge that differences could exist, uncertainty with respect to the ontogenetic age of sampled

teeth and small sample sizes prohibited such an analysis. However, that likely does not impact our ability to interpret plant availability in a region, which is the primary focus of our study. The dental calculus samples from historic and prehistoric elephants include specimens housed in museums and private collections. Because they are mainly historic or prehistoric, pre-dating the establishment of fenced parks and reserves, they are assumed to represent free-roaming elephants. The metadata associated with most specimens is often incomplete, but it helps place them in the geographic and historical context (Tables 2 and 3). Among the freeroaming specimens, the historic specimens, dating to the 1800s and early 1900s are SMEH-1, SMEH-2, SMEH-3 and most probably STL-1 Specimens that may be proto-historic or prehistoric of Holocene age are SMEH-1, SMEH-3, STB-1, and CDL-1 The prehistoric elephants include specimens SME-1, SME-2, SME-3, which correspond to samples from the Lower Paleolithic Elandsfontein site, dated between 1 million and 600,000 years before the present (Klein et al., 2007; Braun et al., 2013). When possible their locations were mapped, although with different levels of accuracy (Fig. 5b). Because the specimens of freeroaming elephants are compared with those of the AENP, they have been divided geographically, namely separating those to the east and west of the AENP (Tables 2 and 3). To the east, samples were more likely associated with the summer rainfall area and mostly in vegetation types of the subtropical thicket and savanna (Table 2). To the west, the studied specimens were in the winter rainfall region, and in vegetation types of the fynbos, renosterveld, forest, and to a certain degree subtropical thicket (Table 3).

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

7

Fig. 4. Model of elephant roaming, vegetation types, biomes, and opal phytoliths.

3.3. Phytoliths in South Africa and the Cape Region 3.3.1. General and regional aspects A study of dental calculus for paleoecological reconstruction requires phytolith reference samples from individual species and taxonomic groups of plants, as well as phytolith groups that characterize the different vegetation types and biomes. Thus far in Southern Africa grass phytoliths have been studied in relative detail (Rossouw, 2010; Cordova and Scott, 2010; Cordova, 2013), although some studies report phytoliths found in other graminoids such as Cypearaceae and Restsionacecae (e.g., Cordova, 2013; Esteban et al., 2016), and non-graminoids (e.g., Mercader et al., 2010). Typically phytoliths are divided by the plant group that produces them and by morphotype. Thus, grass phytoliths are classified as short cells, elongates (or long cells), pointy (or trichomes), and bulliforms (Fig. 6a). Of these, the short cells (i.e., GSSC) are the ones used as diagnostic for taxonomic groups within the Poaceae family. Other graminoids such as Restionaceae (restios) and Cyperaceae (seges) have characteristic forms as well (Fig. 6b). Non-graminoids produce different morphotypes (Fig. 6c) among which the most common in Africa that globular (spherical) and some facetates, which are typically produced by trees (Fig. 6c, numbers 19e21). Other morphotypes are widely found in various dicots, mostly woody plants (Fig. 6c., 18 and 22e25). For this study a number of dicots were tested, but no specific morphotype was found associated with a particular group. The tests on plant specimens carried out through this study showed that globular morphotypes produced by Nuxia floribunda (Fig. 6c, 19) were undistinguishable from those produced by Schotia afra; in other cases many dicot plants produced amorphous types often referred

to as blocky morphotypes. Many plants tested for phytoliths in this study did not produce opal phytoliths as was the case of most succulents (e.g., Portulacaria afra, Aloe spp., and Sensieveira hyacinthoides). 3.3.2. Graminoid phytoliths Among the grass opal phytoliths (Fig. 6a) the short cells are the most reliable for linking morphotypes with taxonomic groups at the subfamily level. This serves as a means for differentiating the two groups of photosynthetic pathways, C3 and C4, used as proxies for climate or soil moisture regime. They are often referred to as grass silica short cells (GSSC) and are usually formed in leaves (Rossouw, 2010), although other studies reported certain types in inflorescences and other parts (Novello and Barboni, 2015; Babot et al., 2016). The classification of GSSC used in this study follows closely the one previously published by Cordova and Scott (2010) and Cordova (2013), but has been slightly modified and updated for the present study (Fig. 7; Table 4). This classification is loosely based on the nomenclature proposed by Madella et al. (2005) and uses terms from Piperno (2006) and a number of other studies (Fredlund and Tieszen, 1994; Alexandre et al., 1997; Barboni and Bremond, 2009; Neumann et al., 2009; Rossouw, 2010). As for the other graminoids, this study uses three categories: Cyperaceae, Restionaceae and generic graminoid. Cyperaceae include mainly the papillae and hats (Fig. 6b, 16 and 17). Restionaceae include several forms such as disks, some of which are flat or have papillae, boomerang-shaped and paddle-shaped elongates that can resemble hockey sticks (Figs. 6b,and 10e15). Although some phytoliths of Restionaceae deteriorate quickly in soils or dental assemblages, they still retain certain characteristic forms

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

8

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Fig. 5. The focus area of study. (a) Soil samples tested for phytolith assemblages and represented in Figs. 8 and 9; (b) approximate location of elephant specimens sampled for dental calculus phytoliths (See Tables 1 and 2).

Table 1 Modern elephant specimens in parks. Park

Designation Materials

Pilanesberg National Park & PBE-1 Game Reserve PBE-2 PBE-3

Addo Elephant National Park

ADE-1 ADE-2 ADE-3 ADE-4

Adult, approximately 30 y. o., one molar Sub-adult, approximately 15 y. o., one molar Juvenile, approximately 5 y. o., one molar Young adult, 18e20 y.o., full mandible molars Adult, >30 y.o., upper molars Adult, >30 y.o., lower molars Adult, >30 y.o., lower molars

(Fig. 6b, 14 and 15). The “generic graminoid” category encompasses types that may occur in more than one graminoid family (Poaceae, Restionaceae, and Cyperaceae), usually some elongates and papillae, and morphotypes that cannot be assigned to a specific graminoid family due to deterioration. 3.3.3. Non-graminoid phytoliths The non-graminoid phytoliths are not as well studied as those of graminoids, but they have recently received attention in locations in tropical Africa, where along with grass phytoliths they have become an important proxy for climatic and environmental conditions (Alexandre et al., 1995; Neumann et al., 2009). Those types are directly associated with taxonomic groups such as the Aceraceae (palms) and several tropical plants found elsewhere in Africa, but not present in the study area. Because no comprehensive

taxonomic study of dicot phytoliths exists in the study area, only three basic groups are used here. One group includes globular types (spherical) and the faceted and tree tracheids (Fig. 6c) because of their strong relation with trees. In a separate category other groups such as the polygons, facetates, blocky types, and unclassified shapes have been amalgamated. To determine the relationship between dicots and graminoids, the dicot/graminoid ratio was used, although its accuracy has not been tested in this region.

4. Methods 4.1. Sampling dental calculus The sampling methodology varied depending on the condition and preservation of the tooth. In elephant molars calculus accumulates at the level of the gums on both sides and on the posterior portion of an erupted tooth. After tartar has been identified on the tooth, the first step is a removal of any varnish, if present. Acetone dissolves most varnishes or solvents. A clean cotton swab was soaked in solvent and applied carefully on the surfaces. Cotton fibers do not contaminate the sample, since they do not produce phytoliths. After removal of the varnish, if any, the tooth surface was gently washed with distilled water. Finally, the surface of the tooth was gently washed with distilled water from a squeeze bottle. A small artist's paint brush was used to remove any sediment from the surface during washing. Detergent was not used since it can dissolve some organic particles attached to the tooth. After the sampling surface was dry, dental tools were used to carefully scratch the tartar off the tooth. A small watercolor paintbrush was then used to sweep the tartar powder into a small sterile vial.

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

9

Table 2 Free-roaming elephant specimens east of the Addo Elephant National Park. Area

Designation

Finding location

Material

Source/collection

Albany District-East London

EL-1 EL-2 EL-3 ALE-1 AME-2 AME-3

Gonubi River mouth East London area East London area Port Alfred area Trappe's Valley, Albany district Amathole Mountains region

One One One One One One

Carl Vernon collection Carl Vernon collection Carl Vernon collection Grahamstown, Albany Museum, #5063 King Williams Town, Amathole Museum, AMSA/M 3274. Registered in 1925 King Williams Town, Amathole Museum, MKM 10823

molar molar molar molar molar molar

4.2. Sampling soil surface samples Phytolith assemblages from soil surface samples were collected to provide a regional and local background of phytoliths produced by the regional and local vegetation. The regional soil surface sample consists of 68 samples collected between 2008 and 2012 along several transects. Part of the data was originally published in Cordova (2013), but more recent samples were taken as part of the current study. The local soil surface samples were collected inside the parks, two from the Pilanesberg National Park & Game Reserve (labeled PB) and five samples from the Addo Elephant National Park (labeled ADDO). Metadata, percentages of data used in this study, and raw counts are provided in the supplementary files S1 (regional samples) and S2 (park samples). At each sampling location, a four-by-four meter quadrat was marked on the ground, where coordinates and elevation were recorded. Then, cover percentages of plant groups were recorded. Four or five random pinches of surface soil were collected from the ground within the quadrat and combined as a single sample for that locality. For purposes of comparison with phytolith assemblages from elephant dental calculus, the percent cover of vegetation is summarized in Table 5.

4.3. Laboratory processing The method used here for extracting phytoliths from herbivore dental calculus takes elements from methodologies employed before (i.e., Armitage, 1975; Gobetz and Bozarth, 2001; ScottCummings and Albert, 2007; Cordova and Agenbroad, 2009). The sample was transferred to a small dish and weighed. Then it was transferred carefully into a 15-ml centrifuge tube. Then, 2e5 ml of 35% HCl was added to the sample and the tube was shaken for 5 min using a vortex genie. Subsequently 2e5 ml of distilled water was added and shaken for another 2 min. The tube was then filled up with distilled water and centrifuged at 2800 rpm for 3 min. The

liquid was carefully decanted, and the test tubes filled up with distilled water again, and centrifuged again. This operation (adding water-centrifuge-decant) was repeated a further two times, or until the pH of the solution was neutral. A microscope slip was weighed in a high resolution balance (at least 3 decimals). The sample was transferred from the tube to a microscope coverslip using a pipette. The sample was evenly distributed on the coverslip. Then the coverslip with sample was placed on a hot plate at 50
C until water evaporated. After drying any pieces of enamel were removed using fine tweezers under a magnifier. The coverslip with sample was weighed and the difference with the clean coverslip provided the sample's weight. Then, a drop of Entellan™ mounting medium was placed on a 1  3 cm microscope slide and the coverslip was quickly placed, sample down on the Entellan medium surface, allowing the still unsolidified medium to flow under the full area below the coverslip. When the Entellan medium was fully dry and solid, after about six or seven days, the slide was ready for microscope work.

4.4. Microscopy, counting, analysis, and presentation of data Once the medium was dry, the sample was scanned under a refraction microscope at 400. If necessary, particularly for photography or study of details, magnification should be increased to 1000 in immersion oil. Normally, counts in soil samples are relatively high (Tables S1 and S2), so that they can be represented in percentages. However, phytolith counts in dental calculus, though highly variable, are generally low (Table 6; Table S3). Therefore, they are reported here as counts. Exceptions include samples of modern elephants at the PNPGR and AENP, as well as some specimens from the eastern part of the zone, namely those labeled EL, AME, and ALE (Table 6), which in this study are also represented in percentages. Whether presented in percentages or in total counts, data in graphs and tables are structured under two groups, graminoid and

Table 3 Free-roaming elephant specimens around and west of the Addo Elephant National Park. Area

Designation

Finding location and circumstances

Material

Source/collection

South Coast

SMEH-1 SMEH-2 SMEH-3 SMEH-4 STL-1

Knysna, killed by Pretorius Addo Bush, 1920 (presumably killed by Pretorius) Schoenmakerskop dunes, Port Elizabeth area Addo Bush, 1897 Knysna

Upper and lower molars One molar Upper and lower molars Upper and lower molars One molar

STB-1

Stilbaai dunes

Two lower molars

SAM ZM-15884 SAM ZM-15737 SAM ZM-33420 SAM ZM-02459 University of Stellenbosch, Animal Science Department collection. Stillbaai Museum display

Olifants River Valley (Citrusdal)

CDE-1

Olifants River valley

One molar

Citrusdal Museum display

West Coast

SME-1 SME-2 SME-3

One molar One molar One molar

Iziko South African Museum Iziko South African Museum Iziko South African Museum

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

10

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Fig. 6. Selected phytolith morphotypes. All bars are 20 mm long. (a) Poaceae phytoliths Short cells: 1, rondel; 2, trapeziform sinuate; 3, panicoid bilobate; 4, short saddle; 5, aristidoid, long and narrow shank bilobate. Other: 6, elongate; 7, trichome; 8, cuneiform bulliform cell; and 9, parallepipedal bulliform cell. (b) Other graminoids, Restionaceae, 10 and 11; disks; 12, boomerang; 13; paddle 13 and 15 deteriorated disk and paddle from dental calculus (SME-2). Cyperaceae: 16, papilla; and 17.hat-shaped cell. (c) Dicots, f ¼ fruit, l ¼ leaf: 18, woody plant tracheid, Ocotea bullata (l); 19, decorated globular, Nuxia floribunda (l); 20 and 21, facetates, Pterocelastrus tricuspidatus (l) and Schotia afra (f), respectively; 22 and 23, polygons, Celtis africana (f,l) and Schyzophiton rautenii (l), respectively; 24; favelolate, Rapanea melanophloeos; and 25, psilate round blocky globular and ovates, Pterocelastrus rautenii.

non-graminoid phytoliths. Graminoids include phytolith morphotypes diagnostic of the Poaceae, Cyperaceae and Restionaceae families and those morphotypes that for various reasons can fall within the three families (i.e., generic non-graminoid). Poaceae morphotypes are subdivided into short cells, elongates, pointy and bulliforms (Fig. 6a). In turn, short cells are subdivided into the groups that characterize the different C3 and C4 subfamilies (Fig. 7). Cyperaceae and Restionaceae are divided into some of the basic groups identified in Cordova (2013), which include papillae, hats, disks, paddles and boomerang-shape morphotyes (Fig. 6b). The non-graminoid group (quasi-equivalent with dicot phytoliths) is summarized into three groups: globular (or spherical), woody plant tracheids, facetates, and faveolates, and other including mainly irregular, blocky types and others not associated with a particular type (Fig. 6c).

5. Results 5.1. Soil samples in the focus area The 68 soil surface samples taken from the seven biomes in the focus area were used here both as a regional background for vegetation and as a modern reference (Fig. 5a). They constitute samples of 42 vegetation types distributed as follows: fynbos (28 types, of which 6 are renosterveld types), succulent karoo (3 types), Afrotemperate forest (1 type), Albany thicket (6 types), nama karoo (2 types), grassland (1 type), and savanna (1 type). The reason for the high number of samples and types in the fynbos is because this biome occupies most of the focus area (Fig. 1a) and has a high diversity of vegetation types (See Mucina and Rutherford, 2006). The distribution of modern phytolith assemblages by biome indicates variations in opal phytolith assemblage patterns as shown

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

11

Fig. 7. Grass subfamilies and their corresponding diagnostic grass silica short cells (GSSC). Bars are 20 mm long, unless otherwise indicated.

Table 4 GSSC used in this study. Designations and equivalents in other sources. Morphotype groups

Code

Reference to other names (source at the bottom)

Poaceae taxonomic group

Rondels and trapezoids

5-A, 5-N, 3-C, 3-F

Pooidea, Danthonioideae and Ehrhartoideae

Oblong trapezoid and crenates

4-C, 4-E, 4-K

Conical (1) (Q) Rondels (2) (G) Trapeziform rondels (3) (I) Pyramidal (1) (Q) Long crenates (1) (E) Trapeziform polylobates & trapeziform sinuates Oblongs (3) (I) Stipa-type (1) Bilobate Variant 2 (3) (I) Trapeziform bilobates (4) Reniforms (3) (I) Bilobates Variant 3 (3) (1) Bilobates Variant 2 (3) (I) Other bilobates (1) (I) Bilobates variant-1 (3) (E) Short saddles (5) (E) Saddles Variant 1 (3) (E) A variant of chloridoid saddles (6) (E) Panicoid bilobates (1) BilobatesVariant-2 (3) (I) Quadra-lobates (2) (E) Crosses (3) (E) Plateau saddles (5) (E)

Trapezoid bilobates and reniforms

8-B, 8-C, 8-E 3-G; 8-E (including reniform variant)

Small, round bilobates

10 L 10Z

Long, narrow bilobates Chloridoid saddles

10A 9A 9B

Panicoid bilobates and crosses

10K, 10W 11A

Phragmites-type saddle

9F

Pooideae and Danthonioideae (2) (E)

Danthonioideae, Pooideae and Ehrhartoideae

Ehrhartoideae, Danthonioideae & Pooideae Aristida (Aristidoideae) Chloridoideae

Panicoideae

Typical of Phragmites (Arundinoideae), but found in some Danthonioideae and some Stipagrostis (Aristidoideae)

Designation sources: (1) Fredlund and Tieszen (1994); (2) Madella et al. (2005); (3) Rossouw (2010); (4) Cordova (2011); (5) Lu and Liu (2003); (7) Piperno (2006, p. 31). Degree of equivalence with designation source: (E) exact equivalent; (I) included as a sub-type in this variant or class; (G) general morphotype designation; (Q), quasiequivalent morphotype

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

12

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Table 5 Information on the localities sampled at the Addo Elephant National Park. Sample and location

Percent area covered by different plant groups 5  5 meters

Trees: none Shrubs and small trees: none Forbs and scrub: unidentified, 5% Grasses: Chloridoideae: Eragrostis sp (30%); Panicoideae grasses: Panicum sp (25%) Cymbopogon sp (2%) Barren: 38% PB-2 Trees: none Near water Shrubs and small trees: Unidentified, 1% Forbs: Unidentified, 2% Grasses Chloridoideae: Cynodon dactylon, 40%. Panicoideae: Panicum sp. 5% Unidentified: 5% Barren: 47% ADDO-1 Trees: None Open area near entrance Shrubs and small trees: Portulacaria afra (24%); other (2%) Forbs and scrub: Pentzia (10%), undetermined geophyte (1%); undetermined succulent (8%) Grasses: Chloridoideae: Eragrostis spp. (30%) Panicoideae: Panicum maximum (15%); Eustachys sp. (5%). Barren: 5% ADDO-2 Trees: none Dense thicket area near entrance Shrubs and small trees: Portulacaria afra, 40%; Azima tetracantha, 2%; Rhus sp., 20% Forbs and scrub: Asparagus sp., 3%; Sensieveira hyacinthoides, 3%; other (1%). Grasses: Chloridoideae: Eragrostis spp., 1%. Panicoideae: Panicum maximum, 5%. Barren: 25% ADDO-3 Trees: none Discovery trail Shrubs and small trees: Azima tetracantha, 3%; Cadaba aphylla, 3%; Euclea sp., 30%; Rhus, 5%; Portulacaria afra, 5%; Carissa bispinosa, 30%. Forbs and scrub: Sensieveira hyacinthoides, 4%; Asteraceae herbs, 3%. Grasses: Pooideae: Stipa dregeana, 5% Panicoideae: Panicum maximum, 5%. Barren: 7% ADDO-4 Trees: Schotia afra, 2% Near water Shrubs and small trees: Azima tetracantha, 4%; Carissa bispinosa, 4%; other unidentified, 15%. Forbs and scrub: Pentzia, 3%; Sensieveira hyacinthoides, 1%; other unidentified, 3% Grasses: Chloridoideae: Cynodon dactylon, 6% Panicoidae: Panicum sp., 2% Barren: 60% ADDO-5 Trees: none Zuurkop hilltop Shrubs and small trees: Euclea undulata, 8%; Carissa bispinosa, 2%, Portulacaria afra, 5%; Syderoxylon inerme, 8%; other unidentified, 5%. Forbs and scrub: Asteraceae, 4%; unidentified, 3% Grasses: Danthonioideae: Merxmuellera disticha, 10% Chloridoideae: Eragrostis, 8%; Chloris sp., 10% Panicoidae: Heteropogon contortus, 5%; Panicum sp., 3%; Themeda triandra, 15% Barren: 17% PB-1 Flat area within 500 m from dam

by the diagram in Fig. 8. The most notable aspect is that the samples from the fynbos biome have a percentages ranging from 60 to 80% of Restionaceae along with smaller percentages (no higher than 10%) of morphotypes classified as Cyperaceae and generic graminoids (Fig. 8, left column). This is concordant with the fact that in most fynbos types Restionaceae tend to dominate over grasses (Cordova, 2013). The non-graminoid to graminoid (NG/G) ratio curve (Fig. 8) does not show a clear pattern through the different vegetation type types. The highest values, in the order of 4e10, are in samples of Albany Thicket (i.e., subtropical thicket). In contrast, the lowest NG/ G values are in grassland and savanna biome, where they are usually less than 1. The forest samples, where one would expect a high NG/G, have values no greater than 4. In fact, the NG/G values are similar to those in the fynbos samples. Although there is high variation of NG/G values in the fynbos, as expected, the lowest

(usually <1) are in the grassy fynbos. The distribution of the sum of C3 and C4 diagnostic GSSC (grass silica short cells) shows an interesting pattern along the different rainfall regimes and biomes (Fig. 8). The C3 diagnostic GSSC tend to be more common in those biomes and vegetation types in the winter and all-year rainfall zones, while the C4 is common in the summer rainfall zone. There is, however, no sharp correspondence between C3 and winter rain and C4 and summer rain, except in some samples of the Albany thicket. Nonetheless, those samples deep in the summer rainfall, such as the nama karoo, grassland and savanna have a predominance of C4 diagnostic GSSC. At the other extreme, the fynbos types have a relatively higher incidence of C3 diagnostic GSSC than other biomes, except for the grassy fynbos type, which is located in the all-year rainfall zone with a considerably higher percentage of summer rain (see location of samples 117e120 on Fig. 5a).

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

13

Table 6 Total counts of phytoliths in the dental calculus samples studied. The asterisk (*) indicates the sets used in the analyses of this study. Park elephants (a) and free-roaming elephants. Sample a) PBE-1 PBE-2 PBE-3 ADE-1 ADE-2 ADE-3 ADE-4 b) EL-1 EL-2 EL-3 AME-2 AME-3 ALE-1 SMEH-1 SMEH-2 SMEH-3 SMEH-4 STB-1 STL-1 CD-1 SME-1 SME-2 SME-3

Diagnostic GSSC (*)

Non- diagnostic/ non-determined GSSC

Other graminoid(*)

Non graminoid (*)

Deteriorated and/or non- determined

Total

6 27 6 78 8 98 44

14 9 14 18 4 33 10

26 23 9 8 8 24 53

61 118 72 45 36 66 26

15 2 0 5 12 40 3

107 177 101 149 56 261 136

33 28 12 108 32 42 4 2 1 4 0 2 3 4 2 2

61 30 46 105 72 152 5 13 1 10 2 4 15 14 15 4

119 76 6 100 69 87 6 20 10 34 6 7 9 38 8 10

60 50 98 38 91 169 37 64 18 44 25 63 16 58 41 28

3 3 1 10 3 2 3 11 4 3 4 3 2 30 14 7

276 187 162 361 267 452 55 110 33 95 37 79 73 144 80 51

In Fig. 9, the individual GSSC morphotypes are separated on the basis of form and affiliation with C3 and C4 subfamilies. There is no particularly good indicator of rainfall regime or biome among the C3 diagnostic GSSC morphotypes. As discussed by Cordova (2013) one problem with the C3 grass subfamilies, particularly Ehrhartoideae and Danthonioideae, is that there is no consistent pattern of phytolith morphology except for a few that are sometimes shared with the Poaceae (e.g., rondels) and reniforms (Rossouw, 2010). However, other morphotypes such as trapezoid bilobates and several long (oblongs) are relatively diagnostic to most C3 grasses (Rossouw, 2010; Cordova, 2013). It is necessary, however, to dedicate a more focused study to the phytoliths of the three C3 grass subfamilies. The C4 diagnostic morphotypes are dominated by the chloridoid saddles (e.g., short saddles), the most common of the C4 diagnostic GSSC, shows a strong abundance in the summer rainfall zone. It dominates in biomes such as the nama karoo and the Albany Thicket. The Panicoideae diagnostic GSSC are present in all biomes, except the succulent karoo, but they tend to be more consistent and relatively abundant in the summer rainfall areas, particularly in the samples from the grassland and savanna biomes. The Aristidoideae-diagnostic GSSC appear only sporadically. They characterize the presence of the genus Aristida, which in this part of Southern Africa mainly indicates disturbance, often by overgrazing (Gibbs-Russell et al., 1990). 5.2. Park and reserve elephants and their environment vegetation The soil samples from the Pilanesberg National Park & Game Reserve (PNPGR) were obtained from the area grazed and browsed by elephants. The vegetation surrounding the samples includes an approximately typical array of the vegetation inside the reserve (Table 6). The phytolith assemblages show a dominance of grass phytoliths. The GSSC show a dominance of C4 grasses, particularly those diagnostic to the Chloridoideae grass subfamily (Figs. 10 and 11). The non-graminoids are represented by a variety of forms, including several of the typical tree-phytoliths such as the globular and faveolated morphotypes (Fig. 7b).

The phytolith assemblages in the three sampled elephant dentitions at the PNPGR show a distribution of phytoliths similar to those in the soil samples (Figs. 10 and 11). Specimens PBE-2 and PBE-3 show strikingly similar proportions among the phytolith groups. Specimen PBE-3, presents a slightly different distribution, with larger proportions of graminoids than non-graminoids. Differences between the elephant and the soil samples exist mainly in the distribution of long, pointy and bulliform grass phytoliths. The phytolith assemblages from dental calculus at the AENP tend to have much lower NG/G values (0.24e1.8) than those in the soil samples (0.79e4.45) (Fig. 12a). In both groups the dental and soil samples, the Chloridoideae-diagnostic GSSC dominate (Fig. 12b), which is consistent with the abundance of Chloridoideae (particularly Cynodon dactylon) in the modern vegetation in areas grazed by elephants in the park (Table 5). Although C3 grasses are present in the sampled areas (Table 5) and overall in the Albany Thicket vegetation type (Mucina and Rutherford, 2006), they are practically absent in the dental calculus of the AENP elephants (Fig. 13). 5.3. Free-roaming elephants in the research focus area The dental calculus phytolith counts of the presumed freeroaming specimens vary considerably by region as shown in the counts (Figs. 14 and 15) and percentages (Fig. 16), which was expected since the samples originate from geographic areas with different vegetation types (Fig. 5b). The most remarkable difference exists in the graminoid morphotype groups, where those specimens from geographic areas characterized by fynbos tend to have Restionaceae and Cyperaceae phytoliths (Fig. 15).). Even in the early-to-middle Pleistocene, fynbos vegetation occupied the same areas it does today, as was suggested by isotopic and mesowear studies on several herbivore remains at Elandsfontein (Luyt et al., 2000; Kaiser and Franz-Odendaal, 2004; Stynder, 2009; Braun et al., 2013). Among the Poaceae phytoliths the differences in C3 and C4 diagnostic GSSC are noticeable. Those specimens collected from

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

14

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Fig. 8. Graminoid phytolith percentages (Poaceae, Cyperaceae, Restionaceae and generic graminoid) types and graminoid to non-graminod from modern surface soil samples (Fig. 4a for location) arranged by biome and winter rainfall. The graph to the far left indicates the sum of diagnostic C3 and C4 GSSC percentages of the total number of grass short cells.

areas the winter and all-year rainfall regions, particularly the West Coast, tend to have more C3 than C4 diagnostic GSSC, while those in the east tend to have more C4 diagnostic GSSC (Fig. 15). Nonetheless, this point is difficult to assert because of the relatively low numbers of phytoliths counted in most dental calculus samples (Table 6).

The samples from free-roaming elephants collected in the Albany District and East London surprisingly produced large numbers of phytoliths (Table 6). Therefore, they can be compared with the relatively phytolith-rich samples of the AENP (Fig. 16). Although there are differences in terms of percentages of short cells, elongated, pointy and bulliforms, for the most part the

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

proportions of non-graminoids and graminoids are similar between the free-roaming and park elephants (Fig. 6). The similarities between the samples of park and free-roaming specimens are perhaps the dominance of C4-diagnostic GSSC, and the lack of Restionaceae phytoliths. The free-roaming elephants, however, have more diversity of C4-diagnostic GSSC and a small but relatively larger number of C3-diagnostic GSSC than their park counterparts (Fig. 16).

15

6. Discussion 6.1. The testing of a model The comparison of the dental calculus phytolith assemblages between samples of the Albany District-East London Area (samples EL-1, 2 &3, AME 2 & 3, and ALE-1) and samples from the AENP (ADE 1, 2, 3 & 4) (Fig. 16) can be used to test the model proposed for this

Fig. 9. Percentages of all GSSC distributed by their corresponding subfamily (See morphotypes on Fig. 6).

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

16

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Fig. 10. Pilanesberg National Park Game Reserve phytoliths from teeth and surface soil sample. Counts.

study (Fig. 4). The Albany District-East London Area samples correspond to the free-roaming elephants, while the AENP samples to the park elephants. Geographically both sample sets correspond to the transition between the all-year and summer rainfall regimes, and the area dominated by the subtropical thicket and patches of adjacent savanna, forest, grassland nama karoo and fynbos). This general area is singled out in the circle in Fig. 1a. In both data sets, the phytolith assemblages look at first very similar (Fig. 16). The most important similarities between the two data sets is high amount of short cells over other Poaceae phytoliths (i.e., elongated, pointy, and bulliform), with only one exception (ADE-4), and the predominance of C4-diagnostic GSSC in all samples (Fig. 16). Additionally, both datasets have relatively similar NG/ G ratios both mostly <1, with only two in each set >1 but <2 (Figs. 11a and 14). However, some notable differences exists between the two groups. First, the free-roaming elephant assemblages have higher richness of phytolith groups than those of the AENP (see “number of morphotypes present” column in Fig. 14). The difference is slight, with an average richness of 9 and 7.75 for the freeroaming and AENP groups, respectively (Fig. 14). In parallel with this difference, the assemblages from free-roaming specimens have a higher variety of GSSC morphotypes (see number of morphotypes present in Fig. 15). The difference in is considerable, with averages of 7.83 and 3.75 for free-roaming and park groups, respectively. Another evident difference in the GSSC is the higher number of panicoid bilobates and crosses in relation to chloridoid saddles in the samples of the free-roaming group. This could be because of their location deeper into the summer rainfall area, where Panicoideae are more prominent (Cordova, 2013), but also because of the broad diversity of veld types for grazing in their ranges.

Although the percentages of C3-diagnostic morphotypes are considerably low in both groups (Fig. 16), the free-roaming elephants group presents a higher variety of C3-related morphotypes (Fig. 15). It is possible that the ample range of the free-roaming elephants covers areas with more grass diversity. Interestingly, however, the lack of Restionaceae phytoliths in the free-roaming group suggests that the elephants probably did not roam in areas of fynbos areas in the west. Instead, they may have grazed areas at higher elevations where lower temperatures promote growth of C3 grasses. Despite the small number of samples, it is evident that the modern elephants in the Addo Elephant National Park have less rich phytolith assemblages than the free-roaming, historical elephants of the same region, suggested in the diagram of Fig. 4. Nonetheless, the invisibility of many plant species in the record may be a problem. But because graminoids produce more distinctive phytolith morphotypes, they could be used as the basis for assessing richness among different regional sets of samples. 6.2. Opal phytoliths and the ecology of the Cape elephants Although traditionally seen as a savanna species, Loxodonta africana subsp. africana, is an adaptable herbivore that can subsist in other African biomes with tropical, subtropical, desert, and temperate vegetation (Blanc et al., 2007; Carruthers et al., 2008). This can be seen in the number of historical elephant occurrences in practically all the biomes of South Africa (Klein, 1983; Seydack et al., 2000; Boshoff and Kerley, 2001; Skead et al., 2007). Although not all the biomes are optimal habitat for elephants, many of the locations where they were historically or archaeologically reported had special vegetation characteristics that provided to their survival. In the region of study many occurrences were associated with low

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

17

Fig. 11. Pilanesberg National Park Game Reserve phyttoliths from teeth and surface soil samples. Percentages.

areas (e.g., floodplains) where there was more water and vegetation (Boshoff and Kerley, 2010). Other occurrences are often reported in the Albany thicket and some of the coastal areas with succulent shrubs highly palatable to elephants (Boshoff and Kerley, 2001). The fynbos biome, despite having a number of historic and prehistoric occurrences (Fig. 3), is one area considered marginal within the elephant range (Boshoff and Kerley, 2001). But one problem to study elephant interaction with fynbos is that there are currently no parks with elephants in this part of South Africa, where preferences for plant foods can be observed. Nonetheless, based on the movements and tracking of Knysna elephants, it is known that some elephants come out of the forest graze in the opening areas with fynbos vegetation (Seydack et al., 2000; Milewski, 2002a, 2002b). One study found that elephants preferred a species of Leucadendron (a shrub of the Proteaceae family) and Bobartia (a bulb of the Iridaceae family), and plants

grown after fire, particularly “grasses and grasslike [sic] plants” (Milewski, 2002b: 32). This suggests that Restionaceae, a grasslike plant, could be eaten. But in another publication the same author stated that observations on their fynbos plant preferences did not include restios (Milewski, 2002a). This contrasts with the findings in our study, since several of the calculus samples from elephants found the fynbos area produced Restionaceae phytoliths, including sample SMEH-1 from the Knysna region (Table 1). Other samples that produced Restionaceae phytoliths include the historic specimens from the fynbos areas around Stilbaai (STB-1) and Port Elizabeth (SMEH-4), one specimen from the pre-park Addo Bush (SMEH-2), and the prehistoric specimens from Elandsfontein (SME1, 2 & 3) (Table 3; Fig. 14). Most fynbos types are poor in grasses (Rebelo et al., 2006), a matter that is also reflected in the phytolith assemblages in modern surface soil samples from the fynbos region, in which not only the

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

18

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Fig. 12. Addo Elephant National Park. Phytoliths from teeth and surface soil samples. Counts.

number of Restionaceae phytoliths exceeds the number of Poaceae phytoliths, but also the non-graminoid ratio is usually above 1.5 (Fig. 8). The presence of Restionaceae phytoliths in almost all the dental calculus samples from the south and west coast (Fig. 14) suggests that in their diets elephants in the fynbos areas may have substituted grasses with restios.

6.3. The case of elephants in parks and reserves The results of our study shows differences in the dental calculus across vegetation areas and between free-roaming and modern park elephants, emphasizing the potential for distinguishing elephant habitats or changing ecologies through the study of diets. In addition to direct observation, one frequent method used in parks for monitoring elephant diets is through plant remains in dung. In the AENP, a study of plant remains in elephant dung showed that the summer and winter diets were dominated by grasses and shrubs, respectively, and that the overall grass diet is 36e38% with the rest composed of forbs, succulents, and shrubs (Paley and Kerley, 1998). Interestingly, that observation is not much different than the composition of plant groups in the dental phytolith assemblages of the four AENP specimens, in which the total graminoid (i.e., mainly grass) vary between 40% and 45% of the total number of phytoliths (Fig. 13). However, one has to account that there are plants that are invisible in phytoliths, and that many nongraminoids produce different amounts than grasses.

According to the plant remains in dung, the grass preferentially consumed by the AENP elephants in the AENP dung is Cynodon dactylon (Chloridoideae), which accounted for the 29% of all the plant weight in the dung (Paley and Kerley, 1998, Table 1). In the same tally, the other only grass reported is Panicum deustum with 3.4% of the total plant weight. The dominance of C. dactylon in the dung corresponds with the predominance of chloridoid saddles in the phytolith assemblages from dental calculus, which for the three of the four specimens is 32%, 30%, and 24% (Fig. 16). In contrast, the panicoid bilobates and crosses make up for 1.3% and 0.9% in only two of the four specimens (Fig. 16). This comparison is suggestive of the potential use of this technique for studying habitat preferences of elephants in different sections of a park, or as shown here, between park elephants and their free-roaming ancestors.

6.4. Potentials and limitations for elephant diet reconstruction The study of large herbivore paleodiets is an important aspect of Quaternary paleoecological research because of the broad implications regarding changing climates, plant-herbivore relationships, and even animal behavior. The most widely used method for reconstructing dietary patterns on modern and past animal populations is the application of stable isotopes on dental enamel and other faunal remains. The stable isotopes of carbon provide a general picture of diet, particularly distinguishing C3 from C4 plants (Vogel et al., 1990; Van der Merwe et al., 1998; Cerling et al., 1999;

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

19

Fig. 13. Addo Elephant National Park. Phytoliths from teeth and surface soil samples. Percentages.

Luyt et al., 2000; Hoppe, 2004; Radloff et al., 2010). In tropical forests and savanna environments where the C3/C4 division often corresponds to the tree/grass division this method is greatly useful. However, phytolith research on dental calculus can add information to stable isotope studies by delineating the divisions between the different C4 grass subfamilies. Additionally, in temperate areas it can also add information on the incidence of C3 grasses. One aspect to consider regarding the reconstruction of herbivore paleodiets is that not all phytoliths embedded in dental calculus originate from plants eaten by the animal. Although there are no studies addressing this problem, one still has to assume that

while most phytoliths are incorporated in the plaque from plant foods, some may also enter through indirect pathways such as soil grit mastication and drinking water. But even if this is the case, the signature of each vegetation type may be recorded as most phytoliths acquired via these other pathways are local to a particular habitat. Despite the potential contributions of dental calculus phytoliths to conservation ecology and paleoecology, it is important to acknowledge their limitations. First, not all plants produce diagnostic phytoliths (Piperno, 2006), which means that many plant groups would be invisible in the record. Second, dental calculus

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

20

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

Fig. 14. Free roaming elephants and Addo Elephant NP (ADE) phytolith summaries. Counts.

Fig. 15. Free roaming elephants and Addo Elephant NP (ADE) GSSC. Counts.

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

21

Fig. 16. Free roaming elephants and Addo Elephant NP (ADE) phytolith summaries and GSSC. Percentages.

may not preserve well in all environments, which may mean fewer counts or no data at all. Third, the number of phytoliths could be highly variable between samples, which can be a problem when comparing cohort groups. In this study, the best preservation and numbers were obtained from the reserve elephants at the PNPGR and the AENP because these are from animals that died recently, and in most cases the specimens were kept protected from exposure to the elements. Other well preserved material came from the specimens from the Albany and East London area, which seem to have been wellpreserved since the time of the shooting of animals. Many of the

specimens date to more recent times, probably the 1800s (Carl Vernon, personal communication). Nevertheless, some older specimens do preserve well, as is the case of the samples from the Pleistocene Elandsfontein elephants. One possible approach to dental calculus samples with low numbers of phytoliths can be overcome by combining cohort samples from herds the same age from one particular site. In paleontological sites where death traps exist, it is possible sometimes to look at individuals as a group, as is the case of data from the Hot Springs Site in South Dakota, USA (Cordova and Agenbroad, 2009). In archaeology, kill sites often provide a number of animals

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

22

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23

that can be grouped as a single sample. For example, the Elandsfontein elephants could be grouped as one cohort if they were to be compared with elephants of the same age in another site. 7. Conclusions Despite the small number of samples and phytoliths in some samples, the study of dental calculus phytoliths presented here points to some preliminary information about elephant ecology in the Cape Region of South Africa. First, phytolith assemblages in dental calculus and permitted differentiation between elephant species in the fynbos area from those of the subtropical thicketsavanna-grassland-forest mosaic of the east. Second, the results of this research show that free-roaming elephants grazed a wide array of graminoids (C3 and C4 grasses, restios, and sedges). Third, the results also showed that, as expected, park elephants have less rich assemblages of phytoliths than the free-roaming specimens collected in the same region. These results suggest that phytoliths embedded in dental calculus constitute a valuable, yet unexplored record of ecological information, particularly on the extent of the species range and habitats. Finally, the application of dental calculus phytoliths may yield important components of diets when undertaken in tandem with direct observations and other dietary studies (e.g., stable isotopes). This approach can be applied to a range of herbivores recovered from arcaheological and paleontological sites, whenever dental calculus is preserved, as well as modern specimens. Although the dental calculus approach was developed for studying elephant paleoenvironnments (including paleodiets) of extinct herbivores it could be applied to long-term studies of wildlife conservation. Acknowledgments We thank the SANParks for permission and support for work at the Addo Elephant National Park; John Adendorf (AENP) and Navashni Govender (Kruger National Park) provided scientific assistance and suggestions. Staff of the Pilanesberg National Park & Game Reserve assisted with the sampling of soils and elephant teeth at the reserve. Staff at Iziko South African Museum are thanked for their support, in particular Denise Hamerton for help locating historic specimens in the museum. We also thank the late Brian Mathiesen  Kotze  of the local of the local museum at Still Bay, and Juandre museum at Citrusdal. Lizette Moolman (SANParks), Professor Graham Kerley and Dr  Boshoff (Center for African Conservation Ecology, Nelson Andre Mandela Metropolitan University) are thanked for their comments and support. Pierre Koekemoer assisted during fieldwork. Finally, we would like to thank the guest editors and the anonymous reviewers for their comments and suggestions. This study was supported by grants from the National Endowment for the Humanities (NEH-ACOR 2006-2007) via the American Center for Oriental Research (Amman, Jordan), and grants from the College of Arts and Sciences, Oklahoma State University (ASR Travel FY 2008). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.quaint.2016.12.042. References Acocks, J.P.H., 1988. Veld Types of South Africa, third ed. In: Memoirs of the Botanical Survey of South Africa 57. Government Printer, Pretoria.

Alexandre, A., Meunier, J.D., Lezine, A.M., Vincens, A., Schwartz, D., 1997. Phytoliths: indicators of grassland dynamics during the late Holocene in intertropical Africa. Palaeogeogr. Palaeoclimatol. Palaeoecol. 136, 213e229. Armitage, P.L., 1975. The extraction and identification of opal phytoliths from the teeth of ungulates. J. Archaeol. Sci. 2, 187e197. Babot, M. d. P., Musaubach, M.G., Plos, A., 2016. An archaeobotanical perspective in the study of inflorescence phytoliths of wild grasses from arid and semi-arid environments of Argentina. Quat. Int. http://dx.doi.org/10.1016/ j.quaint.2016.02.045. Barboni, D., Bremond, L., 2009. Phytoliths of East African Grasses: an assessment of their environmental and taxonomic significance based on floristic data. Rev. Palaeobot. Palynol. 158, 29e41. Blanc, J.J., Barnes, G.C., Dublin, H.T., Thouless, C.R., Douglas-Hamilton, I., Hart, J.A., 2007. African Elephant Status Report: an Update from the African Elephant Database. IUCN-The World Conservation Union, Gland, Switzerland. Blinnikov, M.S., 2005. Phytoliths in plants and soils of the interior Pacific Northwest, USA. Rev. Palaeobot. Palynol. 135 (1), 71e98. Boshoff, A.F., Kerley, G.I.H., 2001. Potential distributions of the medium- to largesized mammals in the Cape Florsitic Region, based on historical accounts and habitat requirements. Afr. Zool. 36, 245e273. Boshoff, A.F., Kerley, G.I.H., 2010. Historical mammal distribution data: how reliable are written records? South Afr. J. Sci. 106, 26e33. Bozarth, S., Hofman, J., 1998. Phytolith analysis of bison teeth calculus and impacta from sites in Kansas and Oklahoma. Curr. Res. Pleistocene 15, 95e96. Braun, D.R., Levin, N.E., Roberts, D., Stynder, D., Forrest, F., Herries, A.I., Matthews, T., Bishop, L., Archer, W., Pickering, R., 2013. Initial investigations of Acheulean hominin behaviour at Elandsfontein. In: Jerardino, A., Malan, A., Braun, D. (Eds.), The Archaeology of the West Coast of South Africa. Archaeopress, Oxford, pp. 10e23. Bremond, L., Alexandre, A., Peyron, O., Guiot, J., 2008. Definition of grassland biomes from phytoliths in West Africa. J. Biogeogr. 35, 2039e2048. Campbell, B.M., 1986. Montane plant communities of fynbos biome. Vegetatio 66, 3e16. Carruthers, J., Boshoff, A., Slotow, R., Biggs, H.C., Avery, G., Matthews, W., 2008. The elephant in South Africa: history and distribution. In: Scholes, R.J., Mennell, K.G. (Eds.), Elephant Management: a Scientific Assessment for South Africa. Wits University Press, Johannesburg, pp. 23e83. Carter, N., 1970. The Elephants of Knysna. Purnell, Cape Town. Cerling, T., Harris, J.M., Leakey, M.G., 1999. Browsing and grazing in elephants: the isotope record of modern fossil prosobsiceans. Oecologia 120, 364e374. Cordova, C.E., 2011. The Stipa-type short cell: what does it mean taxonomically, climatically, and ecologically? In: Scott-Cummings, L., Yost, C., Blinnikov, M., Cordova, C. (Eds.), 8th International Meeting on Phytolith Research. Estes Park, Colorado, Abstracts Book, Golden, Colorado. PaleoResearch Institute, pp. 20e21. Cordova, C.E., 2013. C3 Poaceae and Restionaceae phytoliths as potential proxies for reconstructing winter rainfall in South Africa. Quat. Int. 287, 121e180. Cordova, C.E., Agenbroad, L., 2009. Opal phytoliths from teeth calculus in the mammoths of the Hot Springs Site, South Dakota. Curr. Res. Pleistocene 26, 145e147. Cordova, C.E., Scott, L., 2010. The potential of Poaceae, Cyperaceae, and Restionaceae phytoliths to reflect past environmental conditions in South Africa. In: Runge, J. (Ed.), African Palaeoenvironmental and Geomorphic Landscape Evolution, Palaeoecology of Africa 30. CRC Press Taylor and Francis Group, Boca Raton, Florida, pp. 107e133. Cordova, C.E., Johnson, W.C., Mandel, R.D., Palmer, M.W., 2011. Late Quaternary environmental change inferred from phytoliths and other soil-related proxies: case studies from the central and southern Great Plains. Catena 85, 87e108. Ebedes, H., Vernon, C., Grundling, I., 1995. Past, present and future distribution of elephants in southern Africa. In: van Heerden, J., Penzhorn, B.L. (Eds.), Proceedings of a Symposium on the African Elephant as a Game Ranch Animal. South African Veterinary Association Group, Berg-en-Dal, Kruger National Park, pp. 3e23. Esler, K.J., Pierce, S.M., de Villiers, C., 2014. Fynbos: Ecology and Management. Briza, Pretoria. Esteban, I., De Vynck, J.C., Singels, E., Vlok, J., Marean, C.W., Cowling, R.M., Fisher, E.C., Cabanes, D., Albert, R.M., 2016. Modern soil phytolith assemblages used as proxies for paleoscape reconstruction on the south coast of South Africa. Quat. Int. http://dx.doi.org/10.1016/j.quaint.2016.01.037. Fredlund, G.G., Tieszen, L.T., 1994. Modern phytolith assemblages from the North American great plains. J. Biogeogr. 21, 321e335. Garaï, M.E., Slotow, R., Carr, R.D., Reilly, B., 2004. Elephant reintroductions to small fenced reserves in South Africa. Pachyderm 37, 28e36. Gibbs-Russell, G.E., Watson, L., Koekemoer, M., Smook, L., Barker, N.P., Anderson, H.M., Dallwitz, M.J., 1990. Grasses of Southern Africa. National Botanical Institute, Pretoria. Gobetz, K.E., Bozarth, S.R., 2001. Implications for Late Pleistocene mastodon diet from opal phytoliths in tooth calculus. Quat. Res. 55, 115e122. Golyeva, A., 2007. Various phytolith types as bearers of different kinds of ecological information. In: Madella, M., Debora, Z. (Eds.), Plant, People and Places e Recent Studies in Phytolith Analysis. Oxbow, Oxford, pp. 204e208. Greig, J.C., 1982. The decline and fall of the Knysna elephant. Afr. Wildl. 366, 199e206. Guelke, L., 1989. Freehold farmers and frontier settlers, 1657-1780. In: Elphick, R., Giliomee, H. (Eds.), The Shaping of South African Society, 1652-1840. Wesleyan University Press, Middletown, Connecticut, pp. 66e108.

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042

C. Cordova, G. Avery / Quaternary International xxx (2017) 1e23 Hall-Martin, A.J., 1992. Distribution and status of the African elephant Loxodonta africana in South Africa, 1652-1992. Koedoe 35, 65e88. Hoare, D.B., Mucina, L., Rutherford, M.C., Vlok, J.H.J., Euston-Brown, D.I.W., Parlmer, A.R., Powrie, L.W., Lechmere,-Oertel, R.G., Proches, S.M., Dold, A.P., Ward, R.A., 2006. Albany thicket biome. In: Mucina, L., Rutherford, M.C. (Eds.), The Vegetation of South Africa, Lesotho, and Swaziland, Strelitzia 19. South African National Biodiversity Institute, Pretoria, pp. 541e567. Hoffman, M.T., 1993. Major P.J. Pretorius and the decimation of the Addo elephant herd in 1919-1920. Important reassessments. Koedoe 36, 23e44. Hoppe, K.A., 2004. Late Pleistocene herd mammoth structure, migration patterns, and Clovis hunting strategies inferred from isotopic analyses of multiple death assemblages. Paleobiology 30, 129e145. Kaiser, T.M., Franz-Odendaal, T.A., 2004. A mixed-feeding Equus species from the Middle Pleistocene of South Africa. Quat. Res. 62, 316e323. Kerley, G.I., Landman, M., 2006. The impacts of elephants on biodiversity in the eastern Cape subtropical thickets. South Afr. J. Sci. 102, 395e402. Kerley, G.I., Knight, M.H., De Kock, M., 1995. Desertification of subtropical thicket in the Eastern Cape, South Africa: are there alternatives? Environ. Monit. Assess. 37, 211e230. Klein, R., 1983. Paleoenvironmental implications of Quaternary large mammals in the fynbos region. South African National Science Programme Report 25. In: Deacon, H.J., Hendey, Q.B., Lambrechts, J.J.N. (Eds.), Fynbos Paleoecology: a Preliminary Synthesis. CSIR, Pretoria, pp. 116e138. Klein, R.L., Geis, J.W., 1978. Biogenic silica in the pinaceae. Soil Sci. 126, 145e156. Klein, R.G., Avery, G., Cruz-Uribe, K., Steele, T.E., 2007. The mammalian fauna associated with an archaic hominin skull cap and Later Acheulean artefacts at Elandsfontein, Western Cape Province, South Africa. J. Hum. Evol. 52, 164e186. Kondo, R., Tsumida, T., 1978. The study of opal phytoliths of tree leaves (Part 1): opal phytoliths in gymonsperm and monocotyledon angiosperm tree leaves. Jpn. Soc. Soil Sci. Plant Nutr. 49, 138e144. Lechmere-Oertel, R.G., Kerley, G.I.H., Cowling, R.M., 2005. Patterns and implications of transformation in semi-arid succulent thicket, South Africa. J. Arid Environ. 62, 459e474. Lu, H., Liu, K.B., 2003. Phytoliths of common grasses in the coastal environments of southeastern USA. Estuar. Coast. Shelf Sci. 58, 587e600. Lu, H.Y., Wu, N.Q., Yang, X.D., Jiang, H., Liu, K.B., Liu, T.S., 2006. Phytoliths as quantitative indicators for the reconstruction of past environmental conditions in China I: phytolithbased transfer functions. Quat. Sci. Rev. 25, 945e959. Lu, H.Y., Wu, N.Q., Liu, K.B., Jiang, H., Liu, T.S., 2007. Phytoliths as quantitative indicators for the reconstruction of past environmental conditions in China II: palaeoenvironmental reconstruction in the Loess Plateau. Quat. Sci. Rev. 26, 759e772. Luyt, J., Lee-Thorp, J.A., Avery, G., 2000. New light on the middle Pleistocene West Coast environments from Elandsfontein, western Cape Province, South Africa. South Afr. J. Sci. 96, 399e402. Madella, M., Alexandre, A., Ball, T., 2005. International code for phytolith nomenclature 1.0. Ann. Bot. 96, 253e260. Maud, R.R., 2012. Macroscale geomorphic evolution. In: Holmes, P., Meadows, M. (Eds.), Southern African Goemorphology:Recent Trends and New Directions. Sun Press, Bloemfontein, South Africa. Mercader, J., Astudillo, F., Barkworth, M., Bennett, T., Esselmont, C., Kinyanjui, R., Grossman, D.L., Simpson, S., Walde, D., 2010. Poaceae phytoliths from the niassa rift, Mozambique. J. Archaeol. Sci. 37, 1953e1967. Middleton, W.D., Rovner, I., 1994. Extraction of opal phytoliths from herbivore dental calculus. J. Archaeol. Sci. 21, 469e473. Milewski, A.V., 2002a. Elephant and fynbos. Veld Flora 88, 28. Milewski, A.V., 2002b. Elephant diet at the edge of the fynbos biome, South Africa. Pachyderm 32, 29e38. Mucina, L., Rutherford, M.C., 2006. The Vegetation of South Africa, Lesotho, and Swaziland, Strelitzia 19. South African National Biodiversity Institute, Pretoria.

23

Neumann, K., Fahmy, A., Lespez, L., Ballouche, A., Huysecom, E., 2009. The early Holocene palaeoenvironment of ounjougou (Mali): phytoliths in a multiproxy context. Palaeogeogr. Palaeoclimatol. Palaeoecol. 276, 87e106. Novello, A., Barboni, D., 2015. Grass inflorescence phytoliths of useful species and wild cereals from sub-Saharan Africa. J. Archaeol. Sci. 59, 10e22. Paley, R.G.T., Kerley, G.I.H., 1998. The winter diet of elephant in easter Cape subtropical thicket, Addo elephant national park. Koedoe 41, 37e95. Partridge, T.C., 1998. Of diamonds, dinosaurs and diastrophism: 150 million years of landscape evolution in southern Africa. South Afr. J. Geol. 101, 167e184. Piperno, D.R., 2006. Phytoliths: a Comprehensive Guide for Archaeologists and Paleoecologists. Altamira Press, Lanham, MD. €mberg, C.A., Alimohammadian, H., Sahni, A., 2005. Dinosaur coproPrasad, V., Stro lites and the early evolution of grasses and grazers. Science 310, 1177e1180. Radloff, F.F.T., Mucina, L., Bond, W.J., le Roux, P.J., 2010. Strontium isotope analyses of large herbivore habitat use in the Cape Fynbos region of South Africa. Oecologia 164, 567e578. Rebelo, A.G., Boucher, C., Helme, N., Mucina, L., Rutherford, M.C., 2006. Fynbos biome. In: Mucina, L., Rutherford, M.C. (Eds.), The Vegetation of South Africa,, Lesotho, and Swaziland, Strelitzia 19. South African National Biodiversity Institute, Pretoria, pp. 53e219. Rossouw, L., 2010. The Application of Fossil Grass-phytolith Analysis in the Reconstruction of Cainozoic Environments in the South African Interior. PhD Dissertation. University of the Free State, Bloemfontein. Runge, F., 1999. The opal phytolith inventory of soils in central Africadquantities, shapes, classification, and spectra. Rev. Palaeobot. Palynol. 107, 23e53. Scott, L., 2002. Grassland development under glacial and interglacial conditions in southern Africa: review of pollen, phytoliths, and isotope evidence, Palaeogeography, Palaeoclimatology. Paleoecology 177, 47e57. Scott-Cummings, L., Albert, R.M., 2007. Phytolith and starch analysis of Dent Site mammoth teeth calculus. In: Brunswig, R.H., Pitblado, B.L. (Eds.), Frontiers in Colorado Paleoindian Archaeology: from the Dent Site to the Rocky Mountains. University Press of Colorado, pp. 185e192. Seydack, A.H., Vermeulen, C., Huisamen, J., 2000. Habitat quality and the decline of an African elephant population: implications for conservation. South Afr. J. Wildl. Res. 30, 34e42. Skead, C.J., Boshoff, A., Kerley, G.I.H., Lloyd, P., 2007. Historical Incidence of the Larger Land Mammals in the Broader Eastern Cape. Centre for African Conservation Ecology, Nelson Mandela Metropolitan University, Port Elizabeth, South Africa. Skead, C.J., Boshoff, A., Kerley, G.I.H., Lloyd, P., 2011. Historical Incidence of the Larger Land Mammals in the Broader Northern and Western Cape. Centre for African Conservation Ecology, Nelson Mandela Metropolitan University, Port Elizabeth, South Africa. €mberg, C.A., 2004. Using phytolith assemblages to reconstruct the origin and Stro spread of grass-dominated habitats in the great plains of North America during the late Eocene to early Miocene. Palaeogeogr. Palaeoclimatol. Palaeoecol. 207, 239e275. Stynder, D.D., 2009. The diets of ungulates from the hominid fossil-bearing site of Elandsfontein, Western Cape, South Africa. Quat. Res. 71, 62e70. Twiss, P.C., Suess, E., Smith, R.M., 1969. Morphological classification of grass phytoliths. Soil Sci. Soc. Am. Proc. 33, 109e115. Van der Merwe, N.J., Lee-Thorp, J.A., Bell, R.H.V., 1988. Carbon isotopes as indicators of elephant diets in African environments. Afr. J. Ecol. 26, 163e172. Vlok, J.H.J., Euston-Brown, D.I.W., Cowling, R.M., Hoffman, M.T., 2003. Acocks' Valley Bushveld 50 years on: new perspectives on the delimitation, characterisation and origin of subtropical thicket vegetation. South Afr. J. Bot. 69, 27e51. Vogel, J.C., Eglington, B., Auret, J.M., 1990. Isotope fingerprints in elephant bone and ivory. Nature 346, 747e749. Van Wyck, A.E., Smith, G.F., 2001. Regions of Floristic Endemism in Southern Africa. Umdaus Press, Hatfield, South Africa.

Please cite this article in press as: Cordova, C., Avery, G., African savanna elephants and their vegetation associations in the Cape Region, South Africa: Opal phytoliths from dental calculus on prehistoric, historic and reserve elephants, Quaternary International (2017), http://dx.doi.org/ 10.1016/j.quaint.2016.12.042