ENZOOTIC GEOPHAGY BY ELEPHANTS (LOXODONTA AFRICANA) IN RELATION TO GEOCHEMICAL COMPOSITION OF MINERAL LICKS IN ADDO ELEPHANT NATIONAL PARK, SOUTH AFRICA by Kristen Nadine Darker Thesis submitted in fulfilment of the requirements for the degree Master of Science majoring in Zoology in the Department of Zoology and Entomology, Faculty of Natural and Agricultural Sciences at the University of the Free State South Africa Supervisor: H.J.B. Butler August 2022 ii Declaration I declare that the thesis submitted by me, K.N. Darker, in fulfilment of the requirements for the degree in Master of Science in Zoology at the University of the Free State, is my own independent work. This thesis has not previously been submitted by me or anyone else at another university or faculty. I furthermore concede the copyright of this thesis in favour of the University of the Free State. ______________ __________________ K.N. Darker Date iii Abstract Geophagy, the deliberate ingestion of soil, is a common occurrence amongst various animal species including mammalian herbivores such as elephants. Despite the documented instances of soil-eating, and several nonexclusive hypotheses, the real motivation behind the phenomenon remains controversial. In this study, six camera traps were set up throughout Addo Elephant National Park, South Africa at selected geophagy sites which captured visitation frequency as well as the demographic trend of elephant groups during site visits from April 2019 to May 2020. The geochemical and mineralogical composition of soils at these selected geophagy sites were analysed using X-ray diffractometry (XRD) and X-ray fluorescence spectrometry (XRF). Furthermore, the spatial distribution of five collared elephants (three matriarchs and two males) in relation to the six geophagy sites were investigated using kernel density estimations (KDE). Females had larger home ranges that incorporated more geophagy sites than males. Visitation frequency to geophagy sites were estimated using 500 m buffer zones from the centre of each site. Individuals visited at least three or more geophagy sites throughout the study period. Overall, essential elements Na, Ca and Mn were identified as main drivers for geophagic behaviour in the elephants of AENP. These essential elements (Na, Ca and Mn) are important for certain physiological demands such as bone and tusk growth in elephants and reproductive (pregnancy and lactation) demands in females. Geophagy is considered to be a contributing factor of movement patterns and area utilisation and may have important implications for conservation and management. iv Dedication Dedicated to the parentals. Keith & Linda. Thank you for always being on your knees and being so supportive and invested in my passion. Here’s to breaking generational curses. May knowledge and wisdom, qualifications, and success follow for generations to come. …sodat almal sal sien hoe GROOT is ons GOD. v Acknowledgements To my supervisor, thank you for the amazing opportunity to work alongside these gentle giants. I have learned a lot during this entire process which has shaped me into who I am today. Thank you for the journey, the experiences, the advice and the long nights. We made it! To everyone who was involved in my study at Addo Elephant National Park, a big big thanks. A special mention to Portia Chake. thank you for your assistance in keeping me safe, the long hours, the patience. Charlene Bissett, Lizette Moolman, and Anban Padayachee thank you for the role you played and for making it possible. It was an honour to work with SANParks and their team. To my family and friends who braved the elements and tent living with me, I am ever thankful. Keith, Linda, Brandon, Uncle Marius, Joaquín, Chanel, Natashia, Surene and AJ, I appreciate you all so very much. Another special thanks to my mom and dad for every single contribution to my study, whether it was financial, motivational or simply just a listening ear, it meant the world to me and kept me going. I would also like to extend thanks to Prof. Daryl Codron for statistical advice. Dr Adriaan van der Walt from UFS geography department and Henk Butler, for the GIS assistance and Shaquile Barnes for his talented drawing. As well as Dr Johan Joubert at Shamwari Private Game Reserve for his willingness to help. Lastly, absolutely none of this would have been possible without Dr Joaquín Verdú- Ricoy for whom I have the world’s respect and appreciation. Thank you for being so invested and going above and beyond. Your passion for animals and science is infectious and I have learned so much working with you. I am forever grateful for your time and friendship. vi Table of Contents Declaration Abstract Dedication Acknowledgements Table of Contents List of Figures List of Tables List of Equations List of Appendices List of Abbreviations Ethical Clearance and Permits CHAPTER 1: GENERAL INTRODUCTION 1.1 The concept, causes, and consequences of geophagy 1.2 Elephants as ideal models for geophagy studies 1.3 Aim and objectives of the study 1.4 Thesis outline CHAPTER 2: STUDY AREA AND GEOPHAGY SITES 2.1 Location and topography 2.2 Physiography and geological background of AENP 2.3 Climate 2.4 Vegetation 2.5 Fauna and elephant population history 2.6 Geophagy sites 2.6.1 Identifying geophagy sites in AENP 2.6.2 Description of geophagy sites CHAPTER 3: GEOCHEMISTRY AND MINERALOGY 3.1 Introduction 3.2 Methodology 3.2.1 Soil collection and geochemical analyses ii iii iv v vi vi ii xiii xv xv xvii xviii 1 2 4 8 10 11 12 14 17 18 19 21 21 21 27 28 30 30 vii 3.2.2 Soil classification 3.2.3 Data and statistical analyses 3.3 Results 3.3.1 Soil characteristics and colour classification 3.3.2 Soil mineralogical and geochemical composition 3.4 Discussion CHAPTER 4: VISITATION FREQUENCY AND BEHAVIOURAL ASPECTS OF GEOPHAGY 4.1 Introduction 4.2 Methodology 4.2.1 Camera trapping 4.2.2 Sex classification of elephants 4.2.3 Age classification of elephants 4.2.4 Data and statistical analyses 4.3 Results 4.3.1 Elephant visitation patterns and frequencies 4.3.2 Elephant geophagy by age and sex classes 4.4 Discussion CHAPTER 5: SPATIAL DISTRIBUTION IN RELATION TO GEOPHAGY SITES 5.1 Introduction 5.2 Methodology 5.2.1 Elephant collaring 5.2.2 Data and statistical analyses 5.3 Results 5.4 Discussion CHAPTER 6: GENERAL CONCLUSION 6.1 Summary and general conclusions 6.2 Future studies References Appendix A 56 56 59 63 65 68 68 73 76 84 74 85 88 88 89 91 108 114 115 120 121 142 32 33 34 34 34 47 52 53 viii List of Figures Figure 1 Geographical location of the study area within South Africa (a) and an aerial view of the two sections within Addo Elephant National Park used for this study (b). The blue line indicates a public road that separates the two sections. Pink markers indicate the geophagy site locations. 13 Figure 2 Landform features of the study area in Addo Elephant National Park, South Africa. Green areas indicate high natural vegetation coverage and light brown areas indicate low vegetation coverage. Pink markers indicate the geophagy site locations across the terrain. 14 Figure 3 Climate diagram of Addo Elephant National Park from the period 2005-2020 according to the method of Walter and Lieth (1964). Mean annual temperature and rainfall are indicated in the top left and right corners respectively. The number between brackets indicates years of observation. a) wet season; b) dry season. 18 Figure 4 Classification of dominant soil types according to the method of Batjes (2002) at Addo Elephant National Park, South Africa. 22 Figure 5 Aerial view of the location of each geophagy site (yellow marker) illustrating the surrounding landscape for sites A (a) and B (d). Close-up photographs of geophagy sites A (b-c) and B (e-f) utilised by elephants. 24 Figure 6 Aerial view of the location of each geophagy site (yellow marker) illustrating the surrounding landscape for sites C 25 ix (a) and D (d). Close-up photographs of geophagy sites C (b-c) and D (e-f) utilised by elephants. Figure 7 Aerial view of the location of each geophagy site (yellow marker) illustrating the surrounding landscape for site E (a) and F (d). Close-up photographs of geophagy sites E (b-c) and F (e-f) utilised by elephants. 26 Figure 8 Six of the 18 prepared samples from each of the six geophagy sites showing distinctive colour and hue differences. 31 Figure 9 Principal component analysis (PCA) of major elements (oxides) for all soils sampled at sites A-F. a) loading factors of the first two principal components; b) sites distributed across the first two principal components. 40 Figure 10 Discriminant function analysis (DFA) for major (oxide) elements present at sites A- F, where A-F represent each site respective to corresponding plots (88% of total variance accounted for). 42 Figure 11 Principal component analysis (PCA) of trace elements for all soils sampled at sites A-F. a) loading factors of the first two principal components; b) sites distributed across the first two principal components. 44 Figure 12 Discriminant function analysis (DFA) for major trace elements present at sites A-F, where A-F represent each site respective to corresponding plots (99% of total variance accounted for). 46 x Figure 13 Camera trap photographs taken of different individuals utilising geophagy site B (a), site D (b) and site C (c) respectively. 58 Figure 14 Visual body shape differences between adult bulls (a) and cows (b) as seen from the lateral view (Modified from Moss 1996). 61 Figure 15 Head shape difference between an adult bull (a) and an adult cow (b), as well as the posterior view of an adult bull (c) and an adult cow (d) illustrating body shape and genital differences. (Modified from Moss 1996). 62 Figure 16 Elephant size scaling chart based on average shoulder height and bodily proportions of African elephants. (Modified and redrawn from data published by Henley 2008 and Larramendi 2016). 65 Figure 17 Daily geophagic visits by elephants across all sites throughout the study period. The total number of elephants counts (visits) was standardised (adjusted) in proportion to the season with the highest number of trapping days. 71 Figure 18 Seasonal rainfall (mm) for the duration of the study compared to the seasonal visitation frequency of elephants (n/CTD). The total rainfall for the duration of the study was 445 mm. (Rainfall data obtained from the South African Weather Service for AENP). CTD, camera trap days; n, number of visits to geophagy site. 72 Figure 19 Total proportion (observed values) of elephant age groups across all geophagy sites at Addo Elephant National Park, 73 xi for the entire camera trapping duration from April 2019 to May 2020. Figure 20 Total proportion (observed values) of elephants utilising geophagy sites at Addo Elephant National according to sex, for the entire camera trapping duration from April 2019 to May 2020. Individuals that could not be sexes was marked as ‘unknown’. 74 Figure 21 Total proportion (observed values) of elephants utilising geophagy sites at Addo Elephant National Park, according to the presence or absence of tusks, for the entire camera trapping duration from April 2019 to May 2020. Where the presence or absence could not be determined, it was considered “out of sight” (OOS). 75 Figure 22 Two of the five collared individuals in Addo Elephant National Park, South Africa. (a) matriarch named Bluebell and (b) matriarch named Mushara. 88 Figure 23 Kernel density estimations (KDE) of distribution in relation to geophagy sites (A-F) of Female 1 in AENP. Increase in point density per km2 is presented with a colour gradient to highlight core areas within the home range (yellow line). 94 Figure 24 GPS fixes for collared Female 1 detected within a 500 m radius from the centre of each geophagy site. Each fix is coloured from closest to the site (blue) to furthest from site (red) according to the different buffer distance intervals. 95 Figure 25 Kernel density estimations (KDE) of distribution in relation to geophagy sites (A-F) of Female 2 in AENP. Increase in 97 xii point density per km2 is presented with a colour gradient to highlight core areas within the home range (yellow line). Figure 26 GPS fixes for collared Female 2 detected within a 500 m radius from the centre of each geophagy site. Each fix is coloured from closest to the site (blue) to furthest from site (red) according to the different buffer distance intervals. 98 Figure 27 Kernel density estimations (KDE) of distribution in relation to geophagy sites (A-F) of Female 3 in AENP. Increase in point density per km2 is presented with a colour gradient to highlight core areas within the home range (yellow line). 100 Figure 28 GPS fixes for collared Female 3 detected within a 500 m radius from the centre of each geophagy site. Each fix is coloured from closest to the site (blue) to furthest from site (red) according to the different buffer distance intervals. 101 Figure 29 Kernel density estimations (KDE) of distribution in relation to geophagy sites (A-F) of Male 1 in AENP. Increase in point density per km2 is presented with a colour gradient to highlight core areas within the home range (yellow line). 103 Figure 30 GPS fixes for collared Male 1 detected within a 500 m radius from the centre of each geophagy site. Each fix is coloured from closest to the site (blue) to furthest from site (red) according to the different buffer distance intervals. 104 Figure 31 Kernel density estimations (KDE) of distribution in relation to geophagy sites (A-F) of Male 2 in AENP. Increase in point density per km2 is presented with a colour gradient to highlight core areas within the home range (yellow line). 106 xiii List of Tables Figure 32 GPS fixes for collared Male 2 detected within a 500 m radius from the centre of each geophagy site. Each fix is coloured from closest to the site (blue) to furthest from site (red) according to the different buffer distance intervals. 107 Table 1 Descriptive soil classification of soil per site using colour, mean pH and cation exchange capacity (CEC) values. Soil colour was determined according to the Munsell colour chart and values. 34 Table 2 Identification of essential and non-essential mineral concentrations present at feeding depth soils compared to control soils across all sites A-F in Addo Elephant National Park, South Africa. 36 Table 3 Identification of essential major and trace element concentrations present at feeding depth soils compared to control soils across all sites A-F in Addo Elephant National Park, South Africa. 38 Table 4 Criteria used for age classification and categorising African elephants based on physical characteristics. All calf proportions mentioned are relative to the size and height of an average adult female. (Adapted from Moss 1996, Stokke and du Toit 2000, Whitehouse 2001, Henley 2008). 64 Table 5 Categorisation of the three time periods used for this study that was identified within a 24-hour period per 66 xiv month for Addo Elephant National Park, South Africa. (Data obtained from www.timeanddate.org). Table 6 Seasonal visitation frequencies (n/CTD) of elephants that practice geophagy at six different geophagy sites in Addo Elephant Park, South Africa. CTD, camera trap days; n, number of visits to geophagy site. 70 Table 7 Buffer increment within a 500 m radius set around geophagy sites used to process spatial data obtained from GPS collars for five elephants. 90 Table 8 Number of hourly GPS fixes recorded within a 500 m radius of each geophagy site, for each collared individual. The expected visitation values (italics) were calculated by estimating the % of total GPS fixes for each individual and the multiplying with total number of observed records within the buffer zones (933). The asterix (*) highlights the most visited site for each individual. 92 http://www.timeanddate.org/ xv List of Equations List of Appendices Equation 1 D = √((Lat2 − Lat1) x 110941.85)2 + ((Long2 − Long1) x 111319.44 x Cos(Lat ∗ 2))2 90 Table A1 Mineralogical (X-Ray Diffractometry) analysis and composition (weight %) of primary and secondary minerals present in all soil sampled for sites A-F in Addo Elephant National Park. 142 Table A2 Soil chemistry (X-Ray Fluorescence) of major (in oxide form, weight %) and trace (mg/kg) elements of geophagic soil at site A in Addo Elephant National Park. 143 Table A3 Soil chemistry (X-Ray Fluorescence) of major (in oxide form, weight %) and trace (mg/kg) elements of geophagic soil at site B in Addo Elephant National Park. 144 Table A4 Soil chemistry (X-Ray Fluorescence) of major (in oxide form, weight %) and trace (mg/kg) elements of geophagic soil at site C in Addo Elephant National Park. 145 Table A5 Soil chemistry (X-Ray Fluorescence) of major (in oxide form, weight %) and trace (mg/kg) elements of geophagic soil at site D in Addo Elephant National Park. 146 Table A6 Soil chemistry (X-Ray Fluorescence) of major (in oxide form, weight %) and trace (mg/kg) elements of geophagic soil at site E in Addo Elephant National Park. 147 xvi Table A7 Soil chemistry (X-Ray Fluorescence) of major (in oxide form, weight %) and trace (mg/kg) elements of geophagic soil at site F in Addo Elephant National Park. 148 Table A8 Results of analysis of variance using one-way ANOVAs for normally distributed variables or Kruskal-Wallis (KW) ANOVA (for non-normally distributed variables). 149 Figure A1 Exposed soil layer at site D with scattered shell fragments. (Taken by Kristen Darker). 150 xvii List of Abbreviations AENP Addo Elephant National Park AMC CAM Addo Main Camp Crassulacean Acid Metabolism CEC Cation Exchange Capacity CTD Camera Trap Days DFA Discriminant Function Analysis GI Gastrointestinal GPS Global Positioning Systems KDE KW LOI Kernel Density Estimations Kruskal-Wallis Test Loss on Ignition OOS Out of Sight PCA Principal Component Analysis REE Rare Earth Elements SANParks South African National Parks XRD X-ray Diffractometry XRF X-ray Fluorescence xviii Ethical Clearance and Permits The Interfaculty Animal Ethics Committee of the University of the Free State approved the study project along with the procedures, under the reference UFS-AED2019/0118 on the 18th of July, 2019. Furthermore, research and all data collection in accordance with SANParks regulations were approved under the permit number BUTL-HJB/209-013 issued. 1 CHAPTER 1 GENERAL INTRODUCTION 2 1.1 The concept, causes, and consequences of geophagy The deliberate and frequent consumption of earthy substrates such as clays, soil or sediments, known as geophagy (or geophagia), is a common occurrence among several animal species, including populations of mammalian herbivores and omnivores (Klaus et al. 1998; Chandrajith et al. 2009). Geophagy is hypothesised to be an adaptive, beneficial behaviour in response to a phenomenon known as “pica”, which refers to the craving and intentional consumption of non-food substances (Young et al. 2011; Semel et al. 2019). According to Mahaney and Krishnamani (2003), animals can detect slight variations in the soil composition and would concentrate on a specific area within the same ingested horizon layer. Soil-ingestion among species is encouraged by a physiological need, which may vary in response to seasonal changes, reproductive cycles, and ontogeny, as well as geographical differences, leading to temporal and spatial variation in utilisation (Hui 2004; Blake et al. 2011). Most documented geophagia instances prove that animals are highly selective in the soil substrates they consume (Mahaney and Krishnamani 2003). Geophagy is expected to influence the distribution and densities of wildlife and is crucial for nutritional budgeting (Klaus et al. 1998; Milewski 2000). In the context of animal nutrition and considering the large role minerals and therefore, elements play in geophagy, it is important to clarify the differences in terms used throughout this study. The majority of interdisciplinary literature refers to ‘minerals’ differently and as an all-encompassing term for nutritional elements and/or chemical compounds. However, in this study, I will refer to minerals as naturally occurring chemical compounds, while single chemical elements will be referred to as elements. The mitigation of mineral and elemental deficiencies is frequently mentioned as the immediate cause of geophagy (Weir 1969; Kreulen and Jager 1984; Krishnamani 3 and Mahaney 2000; Stephenson et al. 2011). This theory has gained such widespread acceptance that most locations where geophagy is seen are referred to as "mineral licks," "soil licks," or "natural licks." Numerous research supports the idea that elemental supplementation is the most plausible cause for soil consumption, especially for inorganic elements like sodium (Na), manganese (Mn), potassium (K), and sulfur (S) (Sienne et al. 2014). Despite numerous documented instances of different animal species consuming soil, it is still unclear what the true motivation behind geophagy is. Natural soil consumption has been linked to several potential advantages, including better adsorption of plant phenols and secondary metabolites (Krishnamani and Mahaney 2000), the prevention of acidosis (Klaus et al. 1998), the prevention of diarrhoea (Krishnamani and Mahaney 2000), and the eradication or elimination of endoparasites in the gastrointestinal (GI) system (Kreulen 1985; Klaus et al. 1998; Krishnamani and Mahaney 2000). Alternatively, geophagy could have no physiological benefits whatsoever and the behavioural phenomenon observed could merely be driven by hedonic sensations (Semel et al. 2019). Despite all the proposed advantages, opposing views have also suggested several disadvantages associated with soil eating. Animals that practise geophagy often must travel long distances to reach these sites and this could eventually lead to an energetic disadvantage (Stephenson et al. 2011). Other proposed disadvantages derived from soil eating could result in teeth wearing or increasing the risk of intoxication because of persistent utilisation of contaminated soils (Klaus et al. 1998). In addition, sand impaction can be fatal in case of excessive geophagy (Abutarbush and Petrie 2006). The likelihood of sand impaction is influenced by the animal's daily defecation rate and the quantity of soil it consumes. However, despite possible 4 negative effects, geophagy has been reported from regions all over the world, strongly suggesting that the possible rewards of this behaviour must outweigh the costs. 1.2 Elephants as models for geophagy studies Given that soil eating has a direct effect on the substrate, which can ultimately lead to changes in the terrain of certain areas, geophagy also has important implications for conservation and management (Milewski 2000). This is particularly true for species considered to play an ‘ecosystem engineer’ role in the environment, thus being able to produce significant alterations of their geographical and vegetal characteristics (Valeix et al. 2014). Elephants (Order Proboscidea) are known to practice soil eating as noted in the African savannah (Loxodonta africana), forest (L. cyclotis) and, Asian (Elephas maximus) elephant species across most of their habitats (Spinage 1994). The most striking example of elephants consuming soil is seen in the excavation of ‘caves’ formed by elephants on Mount Elgon's volcanic slopes on the boundary of Kenya and Uganda, as reported by Bowell et al. (1996). Elephants have a diverse diet as they are mixed feeders; they graze primarily in the wet season and browse primarily in the dry season (Mramba et al. 2018). They can dominate in environments which has a significant amount of low-quality plant biomass. According to the Jarman-Bell principle, large-bodied herbivores, such as elephants, can consume plant material with low digestibility and a high concentration of fibre because of their long intestinal retention times and low metabolic demands (Mramba et al. 2018). Geophagy is therefore motivated by GI diseases in elephants that are remedied by ingesting soil (Ayotte et al. 2006; Sienne et al. 2014), or by detoxifying unpalatable foods (Johns and Duquette 1991; Chandrajith et al. 2009). The 5 latter could be the result of a diet based on more browse than grass as various tree and shrub species contain harmful plant secondary metabolites (Mwangi et al. 2004). On the other hand, elephant geophagy has also been hypothesised as a mechanism to obtain mineral and elemental supplementation. According to Cherian (2020), mineral and elemental matter accounts for about 4% of an animal’s body’s weight. For elephants, it is a significant amount as they have an average body weight of 1 800-6 000 kg (Ullrey et al. 1997). A major aspect of fully understanding geophagy is to determine whether animals can discriminate between different elements and minerals in the soil and area. Elephants selectively exploit Na, K, calcium (Ca), and magnesium (Mg)-containing soils as these nutritional elements are typically deficient in their plant diet (Weir 1972). Optimal foraging theories envisage that animals should be able to control diet composition and use foraging tactics that improve energy intake (Ceacero et al. 2010); and due to these necessities of important nutrients, they should seek to ingest additional threshold levels of elements and other essential nutrients. Furthermore, specific nutrient needs (due to seasonality, changing environmental conditions, sex/age classes, and/or reproductive status) can also play a role in elephant geophagy. Elephants have intricate social systems. They frequently form family groups ranging from two to 30 females with kin of varying ages, or males will form smaller bachelor groups made up of adults and sub-adults (Poole 1994). Elephants may, however, form even smaller groups to limit competition when food is scarce (Wittemyer et al. 2005). Bachelor and family groups forage and utilise the habitat differently. Compared to the smaller females with transiently high reproductive demands, males are more tolerant to low-quality forage. Therefore, compared to males that engage in bulk foraging, females are also more selective about the quality of their forage and generally spend more time feeding (Shannon et al. 2006). 6 Ultimately, these large-bodied herbivores' activity patterns and behaviour may be impacted by the difference in the seasonal availability of forage. Holdø et al. (2002) discovered that female elephants in Zimbabwe's Hwange National Park ingested more mouthfuls of soil and expended a larger amount of their daily activity budget ingesting soil than the males. Because Na content in milk and developing foetal tissues is essentially consistent regardless of Na intake, pregnancy and lactation in females impose high demands on Na and other elements (Michell 1995; Blake et al. 2011). Other studies by Kreulen and Jager (1984) and Atwood and Weeks (2003) suggest that Na is the desired element in these soil licks, thus reinforcing the generalised idea about the importance of elephant geophagy on Na intake. However, a preference for salt (NaCl) does not necessarily indicate a deficiency thereof (Phillips 1993). In fact, due to its strong scent, Na might be used by elephants merely as an element-rich site indicator (Sienne et al. 2014). There are numerous other macro- and micronutrients at play in geophagy besides Na. The chemical analysis of geophagy-site soils in the Central African Republic shows significantly higher quantities of K, Ca, Mg, Mn, phosphorous (P) and clay compared to non-geophagic soils (Klaus et al. 1998). In addition, other elements such as chlorine (Cl), and iodine (I) have also been identified as important attractants for geophagy in African forest elephants from Central Africa (Sienne et al. 2014). Although elephant geophagy undoubtedly has some behavioural adaptations, very few research findings on geophagy include the behavioural aspects of the animals involved. Reports of sex- and age differences are usually represented by nothing more than observational remarks. Elephants perform intentional cyclical migrations to geophagy sites in both arid and forest habitats (Panichev et al. 2013), an observation that suggests that seasonal and periodic movement patterns of 7 elephant populations could be determined by the geographical location of geophagy sites. For instance, elephant population density and movement patterns have been proposed to be dependent on the distribution of clay soils high in Na (Weir 1972). Over the past centuries, elephant populations have become increasingly confined to mere fragments of their original distribution range because of anthropogenic influence, a phenomenon that can severely alter their natural behaviour considering they are naturally nomadic over great distances, moving to and from dispersed lick sites (Mwangi et al. 2004). This means that many conservation areas may not, in the long term, be viable for elephant populations unless appropriate provision is made for nutrient supplementation. A study conducted by Leshchinsky (2015), suggests that the mass extinction of woolly mammoths (Mammuthus primigenius), precursors of the modern elephant, was due to geochemical stress caused by mineral and elemental deficiencies. More than 1 500 preserved mammoth bones were analysed, which revealed signs of bone disease such as osteoporosis, osteofibrosis, arthritis and other diseases caused by metabolic disorders due to insufficient essential nutrients. This furthermore emphasises the importance of such resources for elephants. Ultimately ecosystems could collapse without the pivotal role elephants play (Mwangi et al. 2004). 8 1.3 Aim and objectives of the study Within this context, this study aims to explore and describe patterns of geophagy displayed by the African savannah elephant, which is the largest and likely most widely distributed of the elephant species. Understanding potential proximate factors may allow researchers to better understand this iconic species' development and reproductive performance, as well as its migratory patterns and forage quality. To achieve that, six geophagy sites were selected throughout AENP which holds one of the highest densities of elephants in the world (Maciejewski and Kerley 2014). The frequency of elephant visits for geophagy was monitored for approximately 13 consecutive months using camera traps, which also allowed the determination of the age class and sex of the individuals evidencing geophagic activity. Additionally, the chemical and mineralogical composition of soil from the different selected sites was analysed. All these data was obtained to address the following key questions and objectives (divided into specific questions): Key question: What is the geophagy site utilisation pattern and frequency among the elephant population of AENP concerning the geochemical and mineralogical soil properties of selected geophagy sites? Objective 1: To describe the physiography and geology of the study area. Objective 2: To determine the geochemical and mineralogical compositions of soils from each selected geophagy site. 2a. Which major and trace elements and minerals are found in higher concentrations per geophagy site? 9 2b. Do geophagy sites differ in elemental and/or mineral composition from each other? Objective 3: To determine the social structure of the elephants actively utilising the selected geophagy sites. 3a. Is there a difference in the age structure of the elephants using these sites? 3b. Do females utilise these sites more frequently than males? Objective 4: To determine the visitation frequency to geophagy sites among the elephant population. 4a. Is there an increase in geophagy site visits during specific seasons? Objective 5: To determine the spatial distribution of the elephant population in the study area in relation to the selected geophagy sites. 5a. Do their movements patterns include visits to geophagy sites? 10 1.4 Thesis outline This dissertation is composed of six chapters. In this Chapter 1, I have provided a general introduction and background on geophagy, motivation and the significance as well as objectives of this study. The next chapter describes the study area of AENP, its topography and geology, climate and vegetation as well as the history of the Park and its elephant population. Chapter 2 also provides information on each geophagy site utilised and how these sites were identified. Chapters 3, 4 and 5 have been compiled as stand-alone manuscripts to some extent, with each its own introduction, methodology, results and discussion. Due to this, there is some repetition between chapters. Chapter 3 addresses the geochemical and mineralogical aspects of the geophagic study. This includes the methods used in order to obtain geochemical and mineralogical data and presents the results and discussion thereof. Furthermore, Chapter 4 presents the behavioural results obtained from camera traps and discusses the relation to geophagic activity. Chapter 5 reports on the results of Global Positioning System (GPS) collar data provided by South African National Parks (SANParks) to assess the spatial distribution of elephants in relation to geophagy sites in AENP. Finally, Chapter 6 summarises the main findings of this study and discusses the importance thereof for future studies and management efforts. 11 CHAPTER 2 STUDY AREA AND GEOPHAGY SITES 12 2.1 Location and topography The AENP, one of the official 19 national parks (SANParks) of the country, is located east of the Sundays River and approximately 70 km north-east of Gqeberha (formerly Port Elizabeth) in the Eastern Cape Province of South Africa (Fig. 1). The Park also forms the eastern extension of the world-renowned Garden Route. More specifically, this study was conducted in the Addo Main Camp (AMC) and Colchester (excluding the Marine Protected Area) sections of the AENP (Fig. 1). Although the AMC and Colchester sections are separated by the Addo Heights gravel road, together they form one ecological functional unit after the dividing fence was removed in 2006 (Anonymous 2017). Both these sections cover approximately 26 000 ha. The AMC section has the R342 road as its northernmost boundary, while the Colchester section has the N2 highway as its southernmost limit. Topographically, the landscape is characterised by lowland plains and rolling hills (Fig. 2), varying in height between 71 and 354 m above sea level (Paley and Kerley 1998). Numerous natural water pans or small seasonally flooded depressions are distributed across both sections (Toerien 1972; Paley and Kerley 1998). However, permanent water is only available to animals in the AMC via three artificial dams (Hapoor Dam, Rooidam and Domkrag Dam), as well as three water holes (Marion Baree waterhole, Carols’ Rest, and Gwarrie Pan) fed by pumped groundwater (Landman et al. 2012). 13 a Colchester Addo Main Camp (AMC) b Figure 1. Geographical location of the study area within South Africa (a) and an aerial view of the two sections within Addo Elephant National Park used for this study (b). The blue line indicates a public road that separates the two sections. Pink markers indicate the geophagy site locations. Aerial view modified from Google Earth ©2021 in QGIS (Pty) Ltd. 14 2.2 Physiography and geological background of AENP The Greater AENP is situated within the largest onshore basin, an undulating south- dipping fault depression, known as the Algoa Basin (Lombard et al. 2001; Muir et al. 2015). Torrential streams eroded the quartzitic sandstone highlands during the Early Jurassic period and continually deposited extensive granitic boulder beds and gravels (Muir et al. 2017). Along the southern foothills of the Zuurberg Mountain range, a Figure 2. Landform features of the study area in Addo Elephant National Park, South Africa. Green areas indicate high natural vegetation coverage and light brown areas indicate low vegetation coverage. Pink markers indicate the geophagy site locations across the terrain. Modified from Google Earth ©2021 in QGIS (Pty) Ltd. A B C D E D F 15 notable horizon is the resulting red-brown coarse conglomeratic Enon Formation (McLachlan and McMillan 1976). Sedimentation of the Kirkwood Formation started towards the end of the Jurassic Period (ca. 140-150 Mya) and reflects the deposition under prevailing fluvial conditions of fine-grained sediments (Muir et al. 2017). Foliage, petrified wood fossils, and calcrete-rich palaeosols (fossil soils) are present in the resulting multi-hued siltstone, red and green mudstone, and subordinate green-grey sandstone (McLachlan and McMillan 1976; Almond 2014; Muir et al. 2017). The Kirkwood Formation yields the most abundant and diverse animal and plant fossils of all the mid- Mesozoic basin deposits in South Africa (Muir et al. 2015). The Kirkwoord Formation underlies the low-lying plains and the Zuurberg Mountains, which are situated in the south and north, respectively (Kakembo et al. 2015). The Algoa Basin was first filled by fluvial deposition, which was followed by an estuary and shallow marine ingression to create the Sundays River Formation, which is a deposit of mudstone and sandstone that ranges in colour from blue-green to grey (Le Roux 2000). The interlayered, subordinate sandstone bands are often iron (Fe)-rich, thus showing dark brown colour (Toerien 1972). Numerous estuarine to marine megafossils, including ammonites, belemnites, bivalves, gastropods, polychaetes, and echinoids, could be observed in the Sundays River Formation (McLachlan and McMillan 1976; Almond 2014). Molluscs of the Bivalvia (previously known as Lamellibranchiata) class, Trigonia and Exogyra, are extensively spread throughout the Formation (Toerien 1972). The Enon, Kirkwood, and Sundays River Formations make up the Uitenhage Group, which is organised from oldest to youngest (McLachlan and McMillan 1976). The Sundays River Formation sediments underlie the Park's northern portion (which includes the majority of the camp), while the Kirkwood Formation sediments 16 underlie the Park's southern portion (Anonymous 2017). These formations, which are believed to be cyclic transgressive sequences of shallow marine and estuarine origin, have a striking similarity to surface outcrops and often consist of reddish and greenish- grey mudstones, siltstone, and sandstone (Hattingh and Goedhart 1997). Clays with scattered sandstone fragments are prevalent on bare weathered patches and near dams throughout the park. Rocks appear greenish-grey and may contain secondary limestone and gypsum (Hattingh and Goedhart 1997). The current configuration of South Africa's south and east coastlines was formed during the Cretaceous Period (ca. 90 Mya). Following this, subsequent fluctuating sea levels had an impact on the development of the landscape to some extent, exposing the sediments to weathering at times (Toerien 1972). The Coega Platform forms part of the Alexandria Formation near Colchester, as well as the older Grassridge Platform within the AENP near Addo Heights. This formation, which was formed during the Tertiary Period, overlies the Mesozoic Uitenhage Group and is made up of alternating layers of calcareous and quartzitic sandstone, coquinite (cemented shell-rock), and conglomerate with an average thickness of about 9 m (Le Roux 2000). The Alexandria Formation is clearly visible throughout the Park and stands out as a white strip close to the peaks of the hills, especially at the Zuurkop Lookout (Toerien 1972; Anonymous 2006). This white cover layer is also known as the Nanaga Formation, which consists of a dense layer of fine- grained yellowish sandy soils that form the Park's high-lying topography (Almond 2014). The sand and dune rock that make up this layer is also calcareous due to the presence of numerous shell fragments. The Alexandria and Nanaga Formations are two of the five geological groups that make up the Cenozoic Algoa Group. 17 2.3 Climate The AENP is situated within the hot semi-arid region of South Africa (Landman et al. 2008; Kakembo et al. 2015) with a mean annual rainfall of 416 mm and an average temperature recorded of 17.8°C between 2005 and 2020 (Fig. 3). Rainfall is non- seasonal, occurring throughout the year, and it is mostly associated with post-cold frontal events (Hoffman 1989) and usually peaks in austral autumn and spring (Landman et al. 2008). Fogs may provide moisture during extended dry times, as indicated by the prevalence of bark and ground lichens (Barratt and Hall-Martin 1991; Paley and Kerley 1998; Vlok et al. 2003). The maritime and continental climates and the altitudinal variation result in a variable type of climate (Aucamp and Tainton 1984). However, according to Irwin et al. (2008), the Park’s climate is best characterised as warm temperate. 18 2.4 Vegetation Five of the seven biomes found in South Africa are present throughout all the sections of Greater AENP, these being the Nama-Karoo, Fynbos, Albany Thicket, Grassland, and Wetland (Anonymous 2006), making it the most diverse park on the continent concerning vegetation. This wide variety of terrestrial biomes and correspondent vegetation types found throughout AENP is only possible because of the complex 416 mm 17,8 °C mm Figure 3. Climate diagram of Addo Elephant National Park, South Africa, from the period 2005- 2020 according to the method of Walter and Lieth (1964). Mean annual temperature and rainfall are indicated in the top left and right corners respectively. Number between brackets indicates years of observation. (a) wet season; (b) dry season. (Data obtained from South African Weather Service). Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 10 20 30 40 50 60 0 5 10 15 20 25 30 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec P R E C IP IT A T IO N (m m )T E M P E R A T U R E ( °C ) MONTHS TEMP PREC Addo Elephant National Park (15) a a a b b b b Average temperature Average precipitation 19 diversity of soils found throughout the Park (Anonymous 2017). More specifically, the AMC and Colchester sections, that form part of the study area, are situated within the endemic-rich subtropical thicket (sub-order xeric succulent thicket) with high vegetation coverage (vide Fig. 2). It is also the only national park that contains succulent thickets (Landman et al. 2007; Kakembo et al. 2015). The xeric succulent thicket vegetation group is characterised by a dense growth of succulents, evergreen, and spinescent shrubs, lianas reaching between 2-4 m, as well as herbs, geophytes, and grasses (Henley 2001; Vlok et al. 2003; Landman et al. 2007). As a result, the visibility within the thicket is poor (< 5 m) throughout most of the Park (Whitehouse and Schoeman 2003). Dominant shrub and tree species found in the area include spekboom (Portulacaria afra) covering more than 66% of the AMC, needle-bush (Azima tetracantha), wild caper (Capparis sepiaria), white milkwood (Sideroxylon inerme), and plumbago (Plumbago auriculata) (Landman et al. 2007). The grassridge bontveld, which makes up roughly 13% of the region and is found close to the AMC in an area known as Zuurkop, is a combination of woody thickets and small patches of grass veld that are restricted to this highest plateau (Anonymous 2006; Landman et al. 2007; Kakembo et al. 2015). Dominant grass species found include common finger grass (Digitaria eriantha), couch grass (Cynodon dactylon), broad-leaved panicum (Panicum deustum), and red grass (Themeda triandra) (Anonymous 2017; Landman et al. 2007). 2.5 Fauna and elephant population history Initially extending over 2 230 ha, the Park was proclaimed in 1931 to preserve the remaining African savannah elephant population which at the time consisted of only 11 individuals due to extensive ivory hunting locally (Whitehouse and Hall-Martin 2000; 20 Whitehouse 2002; Anonymous 2006). From the 11 survivors, a single male sired about 28% of offspring born over 15 years (Whitehouse and Harley 2002). The population increased exponentially since the mid-1950s after adequate fencing was deployed and is now home to the second-largest elephant population in South Africa, with more than 480 individuals by the year 2008 (Anonymous 2006; Kakembo et al. 2015). By 2020, the elephant population consisted of 492 individuals for the AMC and Colchester sections alone, (Bissett 2021, pers comm*). Elephants constitute about 85% in biomass of the vertebrate herbivores in the Park (Kakembo et al. 2015). Their population has grown beyond the recommended density limit of a carrying capacity of two elephants per km2 (Kerley and Boshoff 1997; Boshoff et al. 2002; Owen-Smith et al. 2006). As the Parks’ boundaries expanded over the years, more species have been gradually introduced into the AENP, including some large browsing and intermediate feeder mammalian herbivores, such as black rhinoceros (Diceros bicornis), greater kudu (Tragelaphus strepsiceros), bushbuck (Tragelaphus scriptus), common eland (Tragelaphus oryx) and red hartebeest (Alcelaphus buselaphus). The bulk of grazing species is comprised of African buffalo (Syncerus caffer) and Burchell’s zebra (Equus quagga burchelli) (Kakembo et al. 2015). In addition, both lion (Panthera leo) and spotted hyena (Crocuta crocuta) were introduced by the end of 2003 into the AMC as part of re-establishing the carnivore process in the Park (Anonymous 2006). The AENP is also home to a wide range of terrestrial birds, herpetofauna, and invertebrates with some species listed as Threatened in the Red Data Book (Anonymous 2006). *Dr. Charlene Bissett, Scientific Services, Addo Elephant National Park, Eastern Cape, 6105, South Africa. 21 2.6 Geophagy sites 2.6.1 Identifying geophagy sites in AENP The geophagy sites for this study were initially identified using Google Earth™, where vegetation clearings could be seen from above, as well as natural animal pathways that lead to the clearing. During the first visit to the Park, each preliminarily identified location was inspected for signs of geophagic activity that indicated active utilisation. Possible geophagy sites reported by the Parks’ officials were also considered and visited to confirm recent utilisation by elephants. Due to the natural dense vegetation found throughout the Park, certain sites could only be reached by foot. Consequently, soil eating could not be directly observed at most of the selected sites, but evidence such as tusk marks, smoothed areas from rubbing, and fresh faecal matter was used. From the initial ten geophagy sites that were considered, six were finally selected for this study, the remaining four being excluded due to several reasons such as inactivity, flooding, erosion, and exposure to park visitors and tourists. Each of the six selected geophagy sites, posed unique landscapes, textures, and soil colours. 2.6.2 Description of geophagy sites The major soil type for the AMC and Colchester sections are luvisols. The dominant soils found throughout the study area are mainly calcic and ferric luvisols (Fig. 4). Luvisols form part of the group of soils that are conditioned by the climate in sub-humid forest and grassland areas and have a well-developed soil structure (Spaargaren 2001; Jones et al. 2013). 22 More specifically, all the geophagy sites, including the areas from where the elephants feed directly, were on vertical and elevated landscapes. They varied from shallow excavations to high exposed soil walls. Soil colours varied from whitish-grey to yellow-brown to reddish or different combinations of these. Due to the physical differences across sites on a smaller scale, it is expected that each site would differ in composition. Alternatively, the sites that are located in the same soil type will be similar in composition. A B C D E F Figure 4. Classification of dominant soil types according to the method of Batjes (2002) at Addo Elephant National Park, South Africa. Derived from the soil database SORTERSAF, ver. 2.1 for southern Africa. Map generated in QGIS (Pty) Ltd. 23 Site A is distinguished by a large east-facing excavation (about 15 m wide) carved into a low-lying hill. This is the only site where the excavation is partly man- made. Excavators are used by park staff to collect soil from this area for use in construction projects throughout the Park, ultimately exposing deeper soil layers. The site is located close to a waterhole and staff housing (Fig. 5). The vegetation in the surrounding area is bare and shows signs of overgrazing. Site B is in a clearing at the top of a slope, roughly 30 m from a tourist road (Fig. 5). The horizontal axis along which soil is actively consumed is approximately 13 m wide. Site C (about 12 m wide) is found along the high walls of a natural deep exposed gulley that forms a natural walkway for animals (Fig. 6). Although the site is very close to a tourist road, it cannot be seen due to the high vertical transects and is accessed by foot. Different soil horizons can be seen along the walls (thick silty to sandy alluvium weathered soils), but geophagy was observed with the focus on the whiter/lighter exposed soils (more calcretised horizon. Site D is located on a hill (Fig. 6). The area where geophagy is observed (about 6 m wide) is on the southern bank of a large depression. The site can only be reached on foot as private roads leading toward the site are overgrown, which also means the site is for the most part undisturbed by human movement. Site E is situated in what appears to be a small dried-up waterhole on a fairly flat landscape and is about 8 m in length (Fig. 7). A few large shrubs are seen near the site; however, most of the surrounding area is rather bare and dry with signs of overgrazing in the area. Lastly, Site F is found on the perimeter of a dried-up dam’s banks. The feeding area about 7 m in length along the edge. It is close to the northern boundary fence of the AMC, as well as train tracks, which are still actively used along the same boundary (Fig. 7). Only on occasion with heavy rainfall, a pool of water will form a few metres from the feeding site. More than one elephant can utilise any of the 6 sites at once. 24 a Figure 5. Aerial view of the location of each geophagy site (yellow marker) illustrating the surrounding landscape for sites A (a) and B (d). Close-up photographs of geophagy sites A (b-c) and B (e-f) utilised by elephants. Aerial views were modified from Google Earth ©2021. d b c e f 25 Figure 6. Aerial view of the location of each geophagy site (yellow marker) illustrating the surrounding landscape for sites C (a) and D (d). Close-up photographs of geophagy sites C (b-c) and D (e-f) utilised by elephants. Aerial views were modified from Google Earth ©2021. a b c d e f 26 Figure 7. Aerial view of the location of each geophagy site (yellow marker) illustrating the surrounding landscape for site E (a) and F (d). Close-up photographs of geophagy sites E (b-c) and F (e-f) utilised by elephants. Aerial views were modified from Google Earth ©2021. a b c D e f d 27 CHAPTER 3 GEOCHEMISTRY AND MINERALOGY 28 3.1 Introduction Geophagy sites or soils (also known as mineral licks or kudurs) are natural landscape complexes that feature mineral/elemental soil outcrops that are deliberately ingested by animals (Young et al. 2011). The majority of geophagy sites fall under the lithomorphic or hydromorphic soil types. While lithomorphic geophagy sites are simply exposed rocks that animals actively search for and ingest, hydromorphic geophagy sites are created by mineral water springs where clay particles become saturated with chemical components in a water discharge region (Panichev et al. 2017). These rich natural soils are typically found in older, more stable landscapes, where they have had adequate time to mature, going through several phases of weathering whereby primary minerals have been converted into secondary (clay) minerals (Mahaney and Krishnamani 2003). Primary minerals are found in sand and coarse silt fractions, whereas secondary minerals can be found in clay and finer silt fractions (Karathanasis 2006). Minerals, and therefore certain elements, are required for a variety of physiological and metabolic processes in animals, including structural growth and support, cellular detoxification, osmotic balance, and immunity (Cherian 2020). Essentially all elements considered in animal nutrition can be divided into two groups: major- and trace elements, depending on the amounts required in the diet rather than their physiological value (Cherian 2020). Geophagia has both beneficial and harmful consequences on animals, depending on the geochemical and mineralogical qualities of the soil consumed (Wilson 2003). Clay deposits, for example, are suggested to aid in the detoxification of the body from anti-nutritional substances by binding noxious compounds or for the relief of GI diseases (Williams et al. 2004; Xu et al. 2004). On the other hand, 29 geophagy could lead to the ingestion of harmful parasites and/or microorganisms or even certain toxic compounds and hazardous metals (Voros et al. 2001). Elephants, among many other species, have reportedly been observed ingesting different soil types that contain a variety of crystal or mineral compositions (Williams and Haydel 2010). Due to limited research on nutritional mineral and elemental requirements in elephants, domestic equids are often used as a reference model considering that both species have a single stomach and a short, but large hindgut fermentation chamber inhabited by bacteria (Clauss et al. 2003). The mineral status of an animal, according to Jansman and te Pas (2015), is the balance between the dietary intake of an element or nutrient and its biological demand. Evidence for mineral and elemental requirements for elephants is very limited. Nutrient requirements could vary due to the age and condition of elephants (Olson 2004). There are several essential elements known to have important metabolic roles in the mammalian system, where dietary deficiency may result in a clinical deficiency. Elements that are considered crucial for biological functionality include Ca, P, Mg, Fe, K, Mn, Na, selenium (Se) and zinc (Zn) as well as trace elements like copper (Cu), cadmium (Cd), arsenic (As), lead (Pb), uranium (U), and vanadium (V) (Sach et al. 2020). Given the complex and diverse scenario that represents the geological base of the AENP, this chapter aims to compare geochemical and mineralogical concentrations across the six different geophagy sites selected for this study. This will be achieved by identifying which of these elements and minerals are more abundant in the XRF and XRD analyses. Determining the mineral and elemental availability in AENP at these selected sites is important as it is thought to shape animals, including elephant distributions, and densities (Milewski 2000). It is predicted the geophagy sites 30 that are located in the same soil type will be similar in composition. The geophagy sites situated in the northern section of the Park will therefore be more similar in composition to each other compared to the sites in the south situated in a different dominant soil type. Alternatively, each site would differ in geochemical and mineralogical composition irrespective of underlain dominant soil types. Furthermore, it is expected that quartz (crystalline silica mineral) will be the found across all sampled sites as it ubiquitous in soils (Jones et al. 2013; Panichev et al. 2013). It is therefore predicted that quartz will not be the mineral sought after instead other minerals/elements present in elevated concentrations. Similarly for trace elements, zirconium (Zr) is also considered ubiquitous in nature (Ghosh et al. 1992) and high concentrations thereof is expected across all sites. Lastly, it is predicted that geophagy would be driven by mineral/elemental supplementation and/or deficiencies due to their physiological nutritional demands. However, geophagy could alternatively or non- exclusively be driven by several GI disorders (vide Chapter 1). 3.2 Methodology 3.2.1 Soil collection and geochemical analyses Soil samples were collected from these six established sites at the end of the study period (August 2020), which allowed sufficient time to confirm geophagic activity. The direct feeding area was established after studying numerous camera trap photographs and confirmed in-field with evidence of active utilisation. Three samples (ca. 150 g each) were collected per site (n = 18), each from the following zones; 1) one sample directly from the feeding or licking area; 2) another sample about 1 m away, along the consumed horizontal axis where possible; 3) one control sample, according to the 31 method described by Mahaney and Krishnamani (2003) and Ayotte et al. (2006). The control sites were randomly selected exposed soil between 8-15 m away from the lick site with no sign of utilisation. Each soil sample collected was stored in a labelled paper bag and left to air dry for several days. All dried samples were subsampled (to approximately 50 g) and ground lightly with a mortar and pestle before being passed through a 2 mm sieve to homogenise the sample for analysis. All subsamples (n = 18) were then stored in plastic containers (Fig. 8) and submitted for semi-quantitative, XRD (X-ray Diffractometry) and quantitative XRF (X-ray Fluorescence), at the Department of Geology of the University of the Free State, Bloemfontein Campus. The XRD technique is commonly used to determine the mineral/crystalline compounds in soil sampled by using the diffraction patterns generated by each unique crystalline phase (Loubser and Verryn 2008). The XRD analysis for this study was Figure 8. Six of the 18 prepared samples from each of the six geophagy sites showing distinctive colour and hue differences. Site A Site B Site F Site D Site C Site E 32 performed using a PANalytical empyrical diffractometer with a Cu-anode ray tube. The software package Highscore/Sleve was used for phase identification and interpretation. Conversely, the XRF technique is used to determine the chemical composition of soil samples. The loss on ignition (LOI) was calculated for major element analysis by heating the samples to 1050 °C and allowing volatile chemicals to escape until their mass ceased changing. The percentage of weight lost on ignition provides an approximate estimation of the soil’s organic composition. Each sample was fused to create beads. For the minor/trace element analysis, all samples were milled using an iron mill before turning the samples into a pressed pellet using Hoechst wax. The Wavelength Dispersive X-ray Fluorescence Spectroscopy (WD-XRF) used for this study was a Rigaku-Primus IV, with a rhodium (Rh) -anode tube, using ZXS software to produce quantitative results. 3.2.2 Soil classification The Munsell colour chart (1994) was used to categorise the colours of the soils, furthermore hue, value, and chroma were used to depict the colours. On the electromagnetic spectrum, the hue notation relates to the colour shade of the soil (R: red, Y: yellow and YR: yellow-red). The purer the colour of the soil, the higher the hue. The value of a soil colour represents how much light is reflected, or how light it is. The concentration of colour is referred to as the soil's chroma. Low-chroma colours are referred to as weak, but high-chroma colours are described as being highly saturated, intense, or vivid. The mean pH (H2O) for three subsampled soils as well as the mean cation exchange capacity (CEC) for three subsampled soil was obtained from soil grids through ISRIC - WDC Soils. 33 3.2.3 Data and statistical analyses Parametric requirements of the (continuous) geochemical variables were first inspected using the Kolmogorov-Smirnov test. LOI was excluded from subsequent analyses. Several approaches were used in the Repeated Measurement Analyses (RMA) using subsamples (feeding sites, 1 m away and control sites) as within site variation. For each mineral/element as cases, only four of 35 differed significantly. Analyses were performed for each combination of geophagy sites and geochemical condition (mineral/crystalline compounds, major- and trace elements) using all the minerals and elements as cases, only two of 18 were significantly different. Using Sites and Type (individually and combined) as categorical predictors showed a non- significant effect of the within-site source of variation. Data from the subsamples were therefor combined in order to get an average value. Consequently, differences in the mean geochemical values among different sites were analysed using one-way ANOVAs (for normally distributed variables) or Kruskal-Wallis ANOVA (for non- normally distributed ones). Two multivariate approaches were used to determine if the geophagy site could be separated according to their geochemical composition. First, a Principal Component Analysis (PCA) was performed for the XRF of major and trace elements of the soil (Treguier et al. 2006), independently. Biplots were used to facilitate the recognition of subgroups to determine the basis of site separation. Additionally, a Discriminant Function Analysis (DFA) was performed (again, independently for the major and trace elements) to investigate how the data were distributed in the morphometric space as done by Hasan et al. 2020, according to their geographic origin (using the site as grouping factor). The standard method was used, as well as the substitution model to include all samples in the analysis. Statistical analyses were 34 performed using STATISTICA 7.0 (Statsoft) with a statistical significance set at p < 0.05. 3.3 Results 3.3.1 Soil characteristics and colour classification The descriptive results of the Munsell soil classification are listed in Table 1 for all sites. The mean soil pH values across all sites varied marginally from 6.1 (slightly acidic) to 6.7 (almost neutral), with sites B, C and D showing the lowest pH values. The CEC ranged between 21-25 cmol (+)/kg. Munsell values differed across all sites but could be grouped according to their hues. Sites A, B and E were redder in colour than sites C, D and F which had a yellow to whitish appearance. Table 1. Descriptive soil classification of soil per site using colour, mean pH and cation exchange capacity (CEC) values. Soil colour was determined according to the Munsell colour chart and values. SITE A SITE B SITE C SITE D SITE E SITE F Soil colour 2.5 YR 5/6 7.5YR 6/6 2.5Y 8/3 5Y 8/2 7.5YR 5/6 2.5Y 7/4 pH (H2O) 6.2 6.1 6.1 6.1 6.7 6.7 CEC (cmol (+)/kg) 23 21 22 22 21 25 3.3.2 Soil mineralogical and geochemical composition The XRD analysis results identified nine different minerals in total: quartz, calcite, plagioclase, muscovite, gypsum, dolomite, ilmenite, K-feldspar/rutile and clinopyroxene (Table 2). In addition, the quantitative XRF analysis yielded the 35 following ten major oxide elements: silicon dioxide (SiO2), titanium dioxide (TiO2), aluminium dioxide (Al2O3), ferric oxide (Fe2O3), magnesium oxide (MgO), manganese monoxide (MnO), calcium oxide (CaO), sodium oxide (Na2O), potassium oxide (K2O) and phosphorus pentoxide (P2O5). The XRF analysis also yielded the following 16 trace elements: chromium (Cr), cobalt (Co), nickel (Ni), scandium (Sc), rubidium (Rb), strontium (Sr), yttrium (Y), Zr, niobium (Nb), barium (Ba), thorium (Th), V, Cu, Zn, As and Pb. The complete XRD analysis results of mineralogical composition found in geophagic soils at sites A-F are presented in Appendix A, Table A1. As predicted, quartz was the most abundant mineral found across all sites except at site D (Appendix A, Table A1), where calcite represented the highest percentage. Quartz concentrations differed significantly among geophagy sites (ANOVA: F5,12 = 18.15, p < 0.001) given the lower concentration of quartz at site D, (mean 32%) compared to the other sites (ranging from 50-73%). The high calcite content (considered abundant, Appendix A, Table A1) of site D (mean 37%) also created significant differences among the sites (K-W ANOVA: H (5) = 13.52, p = 0.019), given the smaller calcite content of sites A, C and E, and its total absence in sites B and F. Overall, plagioclase was the second most abundant mineral found in the soil samples, averaging more than 15% of the total weight, but its abundance did not significantly differ among sites. Muscovite was also common (ca. 10%), but its concentration did show geographical differences (ANOVA: F5,12 = 10.37, p < 0.001), with sites A and E showing slightly higher concentrations than the other sites (Tukey’s HSD, all p < 0.05). The remaining minerals showed residual concentrations (Appendix A, Table A1) in general and were less commonly found among the sites for example, gypsum only being found at sites A and E, dolomite only at sites A and B and clinopyroxene exclusive to site F (5 wt.%). 36 None of the minerals show significant differences among the sampling sites. Table 2 provides a data summary of Appendix A, Table A1 where mineral concentrations detected at the feeding site and the control site were compared to determine which non-essential (quartz) and essential minerals were more abundant. Across all sites, with sites A and B as the exception, quartz concentrations were lower at the direct feeding site. Site C had the most different types of minerals (clinopyroxene, K-feldspar and plagioclase) present at elevated levels in the consumed soil sites as opposed to the control. Dolomite and gypsum were the only two minerals not detected at elevated concentrations across all sites. Table 2. Identification of essential and non-essential mineral concentrations (weight %) present at feeding depth soils compared to control soils across all sites A-F in Addo Elephant National Park, South Africa. MINERALS SITE A SITE B SITE C SITE D SITE E SITE F Calcite X X X Clinopyroxene Dolomite Gypsum Ilmenite X K-feldspar X X X Muscovite Plagioclase X X Quartz O O O O X Essential mineral concentration higher at feeding area than at control O Non-essential (quartz) concentration lower at feeding area than at control Calcite CaCO3 Gypsum CaSO4 · 2H2O Muscovite KAl2(AlSi3O10)(OH)2 Clinopyroxene (Ca,Na,Mg,Fe,Al,Ti)2Si2O6 Ilmenite Fe2+TiO3 Plagioclase NaAlSi3O8/ CaAl2Si2O8 Dolomite CaMg(CO3)2 K-feldspar KAlSi3O8 Quartz SiO2 37 The complete geochemical XRF analysis results are presented in Appendix A, Table A2-A7. The most dominant major element was SiO2 across all sites. However, SiO2 concentration differed among them (ANOVA: F5,12 = 29.57, p < 0.001), with sites B, C and F showing higher SiO2 values than sites A and E, and all of them being higher than site D (Tukey’s HSD, all p < 0.05). Site D also differed from the remaining locations (Appendix A, Table A5) because of its higher CaO concentration (K-W ANOVA: H (5) = 15.37, p = 0.009). RMA showed that Al2O3 was the only major element to differ significantly among the geophagy sites (F2,10 = 10.92, p = 0.003), with sites A and E showing significantly higher Al2O3 concentrations than the remaining four sites (Tukey’s HSD, all p < 0.05). The rest of the major elements showed residual concentrations among the sites. A simplified data summary and of Appendix A, Tables A2 – A7 is provided in Table 3 which highlights the differences of essential major and trace element concentrations detected by the XRF analyses. According to these values, Ca and Na in particular were present at higher or equal levels for most of the sites. Site F on the other hand was the only area containing a higher level of Fe, Mg and As was the most abundant trace elements at higher or equal levels compared to the control, found across four sites. 38 Table 3. Identification of essential major and trace element concentrations present at feeding depth soils compared to control soils across all sites A-F in Addo Elephant National Park, South Africa. SITE A SITE B SITE C SITE D SITE E SITE F M a jo r e le m e n ts Fe X Mg X Mn X X X Ca X O X X X Na X X X X X K P X O X O X T ra c e e le m e n ts V Cu X O X Zn O As X O X X Pb O X X Essential element concentration higher at feeding depth than at control O Essential element concentration remained the same at feeding depth than at control The only significant differences of trace elements across the sites were Ba (F2,10 = 5.6, p 0.023), Rb (F2,10 = 4.79, p = 0.035) and V (F2,10 = 6, p = 0.02). The only exceptions were the level of trace elements at site D, (Appendix A, Table A5) (F2,30 = 5.92, p = 0.007) and C, Appendix A, Table A4, (F2,30 = 4.58, p = 0.018), which were higher for control than for feeding site samples (Tukey’s HSD. Site D: p = 0.005; Site C: p = 0.024). Results from univariate analyses (parametric ANOVA and Kruskal- Wallis ANOVA) are given in Appendix A, Table A8. Overall, these tests evidenced that most of the chemical components analysed (minerals, major and trace elements) showed significant differences in their concentrations among the sites (all p < 0.05), 39 with the only exceptions of plagioclase (ANOVA: F5,12 = 1.73, p = 0.202), ilmenite (K- W ANOVA: H (5) = 9.51, p = 0.091) and dolomite (K- W ANOVA: H (5) = 7.41, p = 0.191). Lastly, as predicted the highest concentration for the trace elements was that of Zr (overall mean 746.6 mg/kg), being the most dominant element across all sites, except for site D (Appendix A, Table A5), where Sr and Ba reached higher levels. Therefore, the Zr concentration was significantly different among sites (K-W ANOVA: H (5) = 14.78, p = 0.01). The second most abundant element was Ba (261 mg/kg), which also showed differences among sites (ANOVA: F5,12 = 41.14, p < 0.001). Tukey’s HSD pairwise comparisons (all p <0.05) showed that sites A (Appendix A, Table A2), E (Appendix A, Table A6) and F (Appendix A, Table A7) (range: 308-367 mg/kg) had significantly higher Ba concentrations than sites D (Appendix A, Table A5), B (Appendix A, Table A3) and C (Appendix A, Table A4) (range: 164-215 mg/kg). The PCA (results of major elements, Fig. 9) determined that about 72% of the total variance was explained by the first two components. The first one accounted for 39% of the variation, with Al2O3, MgO and K2O as the strongest loading factors, all being negatively correlated with the component (Fig. 9a). For the second component (33% of the variance) the strongest loading factors involved SiO2 and CaO, each of them respectively showing negative and positive correlations with the component (Fig. 9a). The first axis differentiated three groups (A+E, C+D and B+F) whereas the second one separated site D from the remaining ones, thus creating a combined effect of four groups in total: A+E, C, D and B+F (Fig. 9b). 40 B1 SiO2 TiO2 Al2O3 Fe2O3 MgO MnO CaO Na2O K2O P2O5 -1,0 -0,5 0,0 0,5 1,0 PC1 : 39.11% -1,0 -0,5 0,0 0,5 1,0 P C 2 : 3 3 .0 4 % A1A2 A3 B1 B2B3 C1 C2 C3 D1 D2 D3 E1 E2 E3 F1 F2F3 -4 -3 -2 -1 0 1 2 3 4 FACTOR 1: 39.11% -3 -2 -1 0 1 2 3 4 5 6 FA C TO R 2 : 3 3 .0 4 % a b Figure 9. Principal component analysis (PCA) of major elements (oxides) for all soils sampled at sites A-F. a) loading factors of the first two principal components; b) sites distributed across the first two principal components. F1 F3 F2 B2 B3 C1 C2 B1 C3 PC1: 39.11% P C 2 : 3 3 .0 4 % FACTOR 1: 39.11% F A C T O R 2 : 3 3 .0 4 % 41 The DFA analysis of the major elements yielded a significant model of discrimination (Wilks’ Lambda: 0.000 approx., F50,17 = 24.703, p < 0.000) with two significant discriminant functions representing 88% of the total variance. The oxides with the largest standardised coefficients for the first function were SiO2 (5.84), Fe2O3 (3.92), CaO (5.33) and K2O (3.49), whereas SiO2 (3.10), Al2O3 (3.85) and K2O (4.99) had the largest standardised coefficient for the second function. Squared Mahalanobis distances among the sites were significant except that between sites B and C, which were the closest in morphometric space (distance = 187.57, p = 0.143, Fig. 10). On the other hand, the highest distance was found between sites B and E (distance = 2162.46, p = 0.002, Fig. 10). The classification matrix generated by the model correctly classified all sites with a 100% correct classification for all soil samples. 42 Figure 10. Discriminant function analysis (DFA) for major (oxide) elements present at sites A- F, where A-F represent each site respective to corresponding plots (88% of total variance accounted for). -30 -20 -10 0 10 20 30 ROOT 1 -20 -15 -10 -5 0 5 10 15 20 25 30 R O O T 2 A C E D F B C F R O O T 2 ROOT 1 43 Considering the trace elements, the PCA analysis results indicate that almost 83% of the total variance was determined by the first two components. The first one (56.7% of the variance) includes Sc, Co, Ni, Zn, Rb, Y and Pb as the strongest loading factors. The second component (26.2% of the variance) included Zr and Nb as the most important loading factors (Fig. 11a). The first function differentiated three groups (A + E, B + C + F and D) whereas the second factor separated sites B and site F from the remaining ones, thus creating a combined effect of five groups in total: sites A + E; B, C, D and F (Fig. 11b). 44 Figure 11. Principal component analysis (PCA) of trace elements for all soils sampled at sites A-F. a) loading factors of the first two principal components; b) sites distributed across the first two principal components. Sc V Cr Co Ni Cu Zn As Rb Sr Y Zr Nb Ba Pb Th -1,0 -0,5 0,0 0,5 1,0 PC1 : 56.70% -1,0 -0,5 0,0 0,5 1,0 PC 2 : 2 6. 24 % A1 A2 A3 B1 B2 B3 C1 C2 C3 D1D2 D3 E1 E2 E3 F1 F2 F3 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 FACTOR 1: 56.70% -3 -2 -1 0 1 2 3 4 5 FA CT O R 2 : 2 6. 24 % a b F1 F2 F3 C2 C1 C3 FACTOR 1: 56.70% PC1: 56.70% P C 2 : 2 6 .2 4 % F A C T O R 2 : 2 6 .2 4 % 45 The DFA analysis of the trace elements (Fig. 12) yielded a significant model of discrimination (Wilks’ Lambda: 0.000 approx., F60,8 = 35.803, p < 0.000) with two significant discriminant functions representing 99% of the total variance. The trace elements with the largest standardised coefficients for the first function were Ba (42.25), Y (38.79), Sr (31.04) and V (33.76), whereas Cr (6.24), Co (5.78) and Pb (5.40) had the largest standardised coefficient for the second function. The longest squared Mahalanobis distance was found between sites D and F (distance = 354957.1, p = 0.013, Fig. 12). The classification matrix generated by the model correctly classified all sites with a 100% correct classification for all soil samples. 46 Figure 12. Discriminant function analysis (DFA) for major trace elements present at sites A-F, where A-F represent each site respective to corresponding plots (99% of total variance accounted for). B D A F E C C C F C ROOT 1 R O O T 2 47 3.4 Discussion The study area in AENP is rich in mudstone topsoil, with underlying layers of clay that are exposed due to weathering of the mudstone. Clay soils can generally be seen over sandstone layers except in heavily weathered areas and open parts of the Park such as the dams where sandstone becomes more visible. All geophagy sites were situated in open vegetation clearings in some instances within close proximity to a water source (e.g., site A). More specifically, the dominant soil type for two geophagy sites in the northern part of the study area (sites F and E) is calcic luvisols. Site D lies on the borderline of two dominant soil types, however, considering the high concentrations of Ca found, it can be deduced that site D also has a more dominant calcic luvisol soil type. The other three sites (A, B, and C) situated in the southern section of the study area, are more dominantly covered with ferric luvisols, therefore, splitting the geophagy sites into two distinct groups based on soil types alone. Luvisols at lower elevations typically have a high percentage of clay, a high capacity to hold nutrients, and a good ability to retain water (Archibald et al. 1996; Jones et al. 2013). It was predicted that the geochemical and mineralogical analyses would support similarities across sites underlain by the same dominant soil types, but the PCA results do not support this prediction. Sites A and E are more similar in composition which is striking as these sites are on opposite sides of the study area and each underlain with different dominant soil types. Soil CEC (mean values) for all sites ranged between 21 and 25 cmol (+)/kg, which has a moderate rating according to Hazelton and Murphy (2007) in terms of the capacity of the soil to hold and exchange cations. Soil pH (mean values) for all sites ranged between 6.1 and 6.7, considered slightly acidic to almost neutral. Site E and F had more neutral mean pH values (both 6.7), and were both located in the north of the 48 Park. Generally, the pH as well as its buffering capacity in relation to CEC affects the availability of nutrients, minerals, and toxic elements in the soil (Hazelton and Murphy 2007). According to Ngole-Jeme and Ekosse (2015), primary minerals that have a lower CEC ultimately influence the sorption capacity of the soils. The high concentrations of the primary non-clay mineral quartz (or forms of silicon oxide) found in this study were expected, as it is abundant in soil environments (Panichev et al. 2013). It is the most common product of natural intermediate weathering of various rocks and due to particles deposited by winds (Jones et al. 2013; Panichev et al. 2013). The high abundance of calcite (CaCO3), which is a secondary (clay) mineral (Ngole-Jeme and Ekosse 2015), found in outlier site D could be accounted for by the presence of numerous calcareous shell-rock fragments (Le Roux 2000; Limpitlaw 2010), that are exposed, as seen in Appendix A, Fig. A1. Half (50%) of the mineral composition at the feeding site at site D consisted of calcite, whereas calcite only made up 37% of the composition for the control sample. This is a notable difference, not only compared to the control sample of site D but also the contrasting low concentration found in sites A, C and E (means ranging between 0.7% and 2,3%). Quartz and plagioclase were the only minerals detected in all samples across all sites. Plagioclase, which was the second most abundant, is a mineral group that consists of a ratio containing Na and Ca and is considered a primary mineral alongside K-feldspar (Hazelton and Murphy 2007), which is found in four (sites B, C, E and F) of the six geophagic soils. According to Taylor and McLennan (1985) and Nesbitt and Markovics (1997), plagioclase is not expected to be present in soils due to its nature to chemically weather and alter rapidly to secondary clay minerals. However, plagioclase feldspars, with respect to europium anomalies (Eu2+), are susceptible to replacing Ca2+ in such plagioclase feldspar matrices (Aubert et al. 2001). For all sites, 49 except site D, plagioclase is found in higher concentrations than calcite while certain sites had no record of calcite at all. As the prediction suggests, concentration ratios of Na and Ca could therefore be elements sought after by elephants across the Park. Other studies conducted claim that Na is an attracting mineral for soil eating (Weir 1972; Holdø et al. 2002; Mwangi et al. 2004). In a study conducted in Aberdares National Park, Kenya, licks that were actively utilised by elephants contained high concentrations of Na and I as they are linked in the geochemical cycle (Mwangi et al. 2004). However, in our study, even though low Na concentrations were present, I was not detected in the soil at all. This is somewhat surprising because according to Johnson (2003), I concentrations in the soil are supposed to be high in coastal zones (0-50 km from the sea) and the furthest site for this study is approximately 33 km from the coastline. Soil’s ability to retain I is depended on soil properties such as soil texture, temperature and pH (Shetaya et al. 2012). Rainfall is considered the main source for I and factors such as duration and intensity effects to what extent I can infiltrate or drain soils (Shetaya et al. 2012). All the above-mentioned factors might have played a role in the draining of I concentrations by the time soil sampling was done and could explain the low levels. This could potentially be alarming, as I is considered an essential element and deficiency thereof could lead to health problems such as thyroid disease (Lu et al. 2005; Shetaya et al. 2012) Furthermore, compared to previous studies on elephant geophagy, such as Houston et al. (2001), no kaolin was detected in the geophagy soils sampled in AENP. According to Chandrajith et al. (2009), an increase in kaolinite (as well as illite) indicates that soils are extremely weathered. Additionally, oxides such as Al2O3 (8.51% mean), CaO (4.41% mean), Fe2O3 (3.36% mean), and K2O (1.24% mean) were found in moderate to low concentrations across all samples compared to the expected high SiO2 (73.21% mean). Al2O3 was 50 the only element with concentrations significantly higher (F2,10 = 10.92, p = 0.003) at the control and 1 m sites than at the feeding site. Ions such as Ca2+, Al3+, K+ and Fe2+ are all exchangeable cations found in the soils. Similarly, it was found that for the surface soils in south-eastern Australia, Fe2O3 and Al2O3, as well as the presence of CaCO3, are dependent factors in the dispersion and flocculation states of soils (Hazelton and Murphy 2007). Obasi and Madukwe (2016) also suggest that silica- alumina (SiO2/Al2O3) ratios serve as an indicator of progressive maturity in sandstone. The presence of Fe2O3 could be explained by the presence of Fe-containing muscovite and ilmenite. Elements such as Fe and Ca, although present in varying quantities, can act as antacids to help relieve excess acidity in the digestive tract (Wilson 2003; Limpitlaw 2010). The values observed for K2O are attributable to the K-feldspar detected in the soil. The higher levels of K could be associated with the natural weathering processes of basement rocks and atmospheric deposition (Nderi et al. 2015). Considering trace elements for the geophagic feeding sites only, the highest concentrations observed were Zr (749.83 mg/kg mean) as predicted, followed by Ba (214.17 mg/kg mean), Sr (101.5 mg/kg mean) and V (78 mg/kg mean). For site B in particular, Zr concentrations were much higher than the rest of the lick sites. Zr, considered one of the ubiquitous rare earth elements (REE), has low solubility and can therefore be used to determine the nature of parent rock and provenance characterisation (Ghosh et al. 1992; Mongelli et al. 2006; Paikaray et al. 2008). However as predicted, by excluding Zr, Ba should be another possible sought after elements across sites. According to Krishnamani et al. (2000), Ba (and Sr) are not of importance for geophagy behaviour. Interestingly, Ba and Sr are found in the calcium- normalized breastmilk of mammals, like walruses, and the concentration found in 51 calves is directly proportional to the bioavailability thereof found in the mothers’ body (Clark et al. 2020). How exactly these elements play a role as possible nutrient drivers for elephants is unknown. Geophagy is hypothesised to be driven by various and complex functional, physiological and morphological changes and/or deficiencies and/or stress. The mineral/elemental concentrations (and combinations thereof) between and within sites might differ on a finer scale, however results suggest the Na and Ca might be the drivers of geophagy behaviour in the AENP elephant population. Elephants require a high Ca and Na intake, especially for tusk growth and when females lactate (Dierenfeld 2008). Ca deficiency and imbalances could lead to metabolic bone disease (Ensley et al. 1994; Leshchinsky 2015). On the other hand, it is worth mentioning that the excessive amounts of quartz and toxic metals such as Cr and Ni in the soils represent possible health threats for geophagic species (Nderi et al. 2015; Ekosse et al. 2017). 52 CHAPTER 4 VISITATION FREQUENCY AND BEHAVIOURAL ASPECTS OF GEOPHAGY 53 4.1 Introduction Elephants have been documented practising geophagy in several previous studies conducted in different parts of the world: Central African Republic (Ruggiero and Fay 1994), Tanzania (Houston et al. 2001; Kalumanga et al. 2016), Zimbabwe (Weir 1972; Holdø et al. 2002), Kenya (Bowell et al. 1996; Mwangi et al. 2004), and Sri Lanka (Chandrajith et al. 2009). For different reasons, elephant species in different habitats will seek out natural soil licks, which are located where nutrients are concentrated (Klaus and Schmid 1998; Mwangi et al. 2004). These mineral and elemental hotspots are important habitat features that play a large role in determining the behaviour within an individual’s home range as well as how the landscape is utilised (Hunter 2017). Geophagy sites may be fixed in space but could only be required at certain periods and therefore may temporally influence their behaviour (Davies et al. 2016). A study conducted in the Aberdares National Park in Kenya by Mwangi et al. (2004), found that elephants ingested soils with higher concentrations of Na and I. Given the limited availability of grasses in Aberdares National Park, the elephants’ diet consists of mainly browse and certain seasonal fruits which generally contain more harmful secondary substances like tannins and alkaloids (Mwangi et al. 2004). On the other hand, in Zimbabwe’s Hwange National Park, a study conducted by Holdø et al. (2002) also found that the consumed soil was Na-rich. However, more female elephants were observed ingesting these soils, and it was concluded that geophagy might be driven by the high physiological demands in elephants during pregnancy and lactation. Another geophagy study found that Asian elephants in Udawalawe National Park, Sri Lanka, consumed kaolinite and illite-rich soils (Chandrajith et al. 2009). It is suggested that these soils aided in the detoxification of unpalatable compounds found in their diet instead of nutrient supplementation (Chandrajith et al. 2009). 54 Several non-exclusive hypotheses exist to explain the role and prevalence of geophagic behaviour and can be divided into two broad categories: protection against gastrointestinal (GI) disorders and nutrient supplementation. Under the protection hypothesis, alleviation of GI disorders can be subdivided into four non-exclusive hypotheses: 1) relief from endoparasites and pathogens in the gut by ingesting adsorptive clays (Krishnamani and Mahaney 2000). 2) Adsorption of plant secondary metabolites and toxins (Pebsworth et al. 2011). Studies by Mwangi et al. (2004) and Chandrajith et al. (2009) have proposed that elephants practice geophagy, particularly for the detoxification of secondary plant metabolites in their vegetation diet. Toxins encountered through their diet include tannins, terpene, alkaloids, and phenols. 3) Ingesting substances like kaolin clays for their adsorptive ability could act as an antidiarrheal agent (Dominy et al. 2004). 4) Antacid effect, as ingestion of clay can assist in the adsorption of organic molecules acting as a buffer in the stomach and gut lining by producing extra mucous secretion (Leonard et al. 1994). However, the more universally accepted hypothesis explaining this habitual and intentional behaviour is that geophagy supplements dietary mineral and elemental deficiencies (e.g., Klaus et al. 1998; Young et al. 2011; Kalumanga et al. 2016; Pebsworth et al. 2019). Dietary changes may modify the demand for geophagy, resulting in seasonal fluctuations in soil lick utilisation (Blake et al. 2011). Elephants are generalist/mixed feeders with a typical diet of grasses and browse, which could generally lack the adequate nutritional element concentrations to meet their dietary demand (Kalumanga et al. 2016). Deficits in micronutrients can have a negative impact on health status and raise the risk of disease (Rode et al. 2003). Animals also need different nutritional elements during different life stages such as reproductivity 55 and lactation (Tracy and McNaughton 1995; Atwood and Weeks 2003) and for bone (Henshaw and Ayeni 1971) and tusk growth (Whitehouse 2002). Elephant visitation or group dynamics, as well as the frequency of visits to these sites, can vary from one lick to the next (Tobler et al. 2009). Such variance could be due to differences in soil composition, distances individuals need to travel, topographical accessibility or perhaps be linked to the presence or absence of predators (Izawa 1993; Tobler et al. 2009). The spatial and temporal heterogeneity of resources across the habitat greatly influences the way elephants utilise the landscape (Weir 1972; Ferreira et al. 2017) as well as the dynamics of social structures and group sizes formed (Poole and Moss 1989). Generally, three types of groups are formed in elephants: 1) All-male/bulls’ group; 2) cows and calves’ groups and 3) mixed groups which consist of bulls, cows, and calves (Moss 1996). A population's age and sex distribution can be affected by factors including calf survival, birth and death rates, conception rates, and other behavioural traits (Dublin and Taylor 1996). Numerous factors influence variation and shifts in group sizes, whether seasonal or long-term, which may have significant implications on the ecology of any ecosystem (Western and Lindsay 1984). Understanding elephant behaviour and population dynamics may therefore have significant effects on management strategies for elephants. It had already been noted by Kerley and Landman (2006) that the African savannah elephants have been observed to consequently alter the landscape in AENP by excavating soil for consumption, yet it had not been investigated. Therefore, this chapter aims to address the patterns of visitation by elephants to selected geophagy licks located in AENP. It is predicted that elephants, being diurnal animals, will utilise geophagy sites more frequently during the day when they are actively foraging. Alternatively, it could be 56 predicted that visitation frequencies would peak during the night, as this is when most predators are least active, the atmospheric temperatures are lower (lower energetic cost for elephants), human interactions are minimal and/or elephants would possibly engage in geophagy after foraging to mitigate possible GI disorders. Another prediction is that seasonal utilisation would yield an increased and consistent pattern during the colder seasons i.e., winter and autumn. This is usually when vegetation and therefor nutrient availability is lower across the landscape. Alternatively, more frequent utilisation during the spring and summer months could support the hypothesis that more nutrients are required for breeding and reproduction. Due to the high nutritional requirements of reproductively mature females (especially Na and Ca) during pregnancy and lactation, it is predicted that adult females will utilise the mineral rich geophagy sites more often. Alternatively, taking into consideration their sexual dimorphism, males need a higher Ca and P intake than females for bone growth. Similarly, it is also expected that more individuals with tusk than without with exhibited geophagy behaviour due to the higher nutritional demand thereof. 4.2 Methodology 4.2.1 Camera trapping The use of camera traps is a popular non-invasive technique for examining species composition, structure and abundance, habitation, density, and activity patterns (Swann et al. 2004; Pimm et al. 2015). The AENP has dense vegetation and most of the identified geophagy sites can only be reached by foot, making direct observations dangerous and difficult for a long period of time. Camera traps were therefore 57 considered the appropriate method to continuously capture geophagic events and visits by the animals of the Park without human interaction and influence. During the first visit to AENP in April 2019, ten initial sites were selected, and a camera trap was deployed at each site at strategic locations depending on the terrain. Considering that geophagy sites were situated within vegetation clearings, camera traps could not be mounted to trees, so modified stands were used instead. Given the limited number of suitable camera traps available, only one camera trap was placed at each geophagy site which adequately covered the geophagy area similar to methods of Pebsworth et al. (2011). More than one elephant could be photographed at once utilising the site at the same time (Fig. 13). Two different types of cameras were used, Bushnell HD Trophy Cameras and Browning Spec Ops Advantage Trail Cameras, both triggered by an infrared motion- and-heat detector. Similar, to the methods of Blake et al. (2011), all cameras were set to take three consecutive photographs once triggered with a minimum of the 5-min time interval between instances. For the whole study period (April 2019-May 2020) the cameras remained continuously active unless technical errors occurred. Most of the interruptions, however, were caused by the elephants displacing the cameras or damaging them, thus greatly affecting the number of successful camera trapping days at certain sites. The nation-wide lockdown during the Covid-pandemic also contributed to missing data. During this period, travel restrictions and strict SANParks regulations prohibited visits in order to replace batteries, empty SD-cards and restore knocked over cameras. Cameras were serviced approximately every 5 weeks (when possible) to exchange batteries and memory cards, as a preliminary test proved that the batteries would last up to 7 weeks with our set up of choice. 58 Figure 13. Camera trap photographs taken of different individuals utilising geophagy site B (a), site D (b) and site C (c), respectively. a A b A c A 59 All photographs were automatically stamped with the date and time once triggered. Based on the 5-min interval setting of the camera, any photograph triggered after that period was considered an independent record, similar to the methods of Srbek-Araujo and Chiarello (2005) and Hamel et al. (2013). Records that included individuals passing by the area, not in cl