Evaluating the Role of Savanna Elephants in the Seed Dispersal Ecology of Lala Palms (Hyphaene coriacea)

Research contextThroughout Earth’s history, megaherbivores, i.e. large-bodied herbivorous vertebrates exceeding one tonne, have been fundamental in shaping terrestrial ecosystems through herbivory, trampling, and long-distance seed dispersal (Janzen & Martin, 1982; Owen-Smith, 1988). These ecological processes have profoundly influenced plant community structure, species distribution, and evolutionary trajectories (Herrera, 1995; Donatti et al., 2007). However, widespread megaherbivore extinctions over the past millennia have led to the loss of many of these ecological functions, especially outside of Africa (Hansen & Galetti, 2009). Africa has a relatively intact megaherbivore assemblage, which offers a rare opportunity to study once widespread ecological dynamics that are now globally rare. The African savanna elephant (Loxodonta africana) is distinguished among extant megaherbivores for its ecological importance and for its continuous evolutionary lineage. As one of three surviving species of the order Proboscidea, which originated in Africa around 60 million years ago (Shoshani, 1998), African elephants are an example of a lineage that has persisted through extensive global megafaunal losses. Africa’s retention of large-bodied frugivores like elephants provides a rare opportunity to investigate ancient plant-animal interactions that have likely co-evolved over millions of years (Kalb, Froehlich & Bell, 1996), particularly those involving megafaunal fruits, whose large, fleshy structures are considered adaptations for dispersal by megafauna (Guimarães Jr, Galetti & Jordano, 2008). While several studies have examined the loss of seed dispersal services following megafaunal extinctions (Janzen & Martin, 1982; Donatti et al., 2007), less attention has been paid to regions where these interactions still persist. In southern Africa, elephants and large-fruited trees coexist across vast savannas, offering insights into functioning plant-animal mutualisms. The ecological role of elephants extends beyond herbivory; elephants facilitate long-distance seed dispersal, and thus probably influence plant population connectivity, gene flow, and regeneration dynamics (Dudley, 2000;Campos-Arceiz & Blake, 2011; Bunney, Bond & Henley, 2017). Despite being recognised as keystone seed dispersers, empirical data quantifying their dispersal services – such as gut passage effects on the seeds and seed dispersal distances – for specific megafaunal fruit species remain limited and need further investigation. Given increasing habitat fragmentation and the decline of elephant populations in parts of their range, understanding the ecological role of savanna elephants as seed dispersers is critical for informing conservation and management strategies. This dissertation addresses this knowledge gap by examining the dispersal ecology of Hyphaene coriacea (Lala palm), a large-fruited species found within the savanna biome of southern Africa. Through controlled feeding trials, germination experiments, and movement-based dispersal modelling, this study aims to generate a comprehensive assessment of the seed dispersal service provided by savanna elephants, situating these processes within broader ecological and evolutionary frameworks.ObjectivesThis study is guided by the following objectives:Objective 1. To determine the gut-passage time of H. coriacea fruits in savanna elephants and to evaluate the extent to which elephant biological traits (specifically body size) influence seed survival during the digestive transit.Objective 2. To assess the effect of gut-passage through savanna elephants on the germination rate and success of H. coriacea seeds in comparison to non-ingested control seeds.Objective 3. To model and estimate the spatial extent of H. coriacea seed dispersal facilitated by savanna elephants, with a focus on quantifying long-distance dispersal events across the landscape. This is done by applying the approach developed by Westcott et al. (2005), combining feeding trial data from sanctuary African savanna elephants with movement data from wild radio-collared individuals.By achieving these objectives, this research aims to contribute to a deeper understanding of the ecological role of savanna elephants as functional seed dispersers and to inform conservation strategies that protect plant-elephant mutualisms within African savanna ecosystems.MethodsTo determine gut passage time of H. coriacea seeds, we intended to record the number of days elapsed between feeding fruits to elephants and retrieving the seeds from the dung. The feeding experiments were conducted on eleven African savanna elephants, seven females and four males, with body sizes ranging between 1.9 and 5.8 tons. For the main feeding trial, fruits were mixed and allocated randomly in two treatments: (1) fruits that were fed to the elephants, and (2) a control treatment, consisting of fruits that were not ingested. On 22 September 2025, we offered 30 fruits (360 in total) to each elephant in one bout. The eleven elephants included in this study are housed across three separate stables. Each stable consistently houses the same group of individuals. Following each dung pile disaggregation, all identifiable remains of lala palm fruits were carefully retrieved and immediately stored in individual paper envelopes matching the exact inspected dung pile. This included intact fruits, intact seeds, fragments of the fruit mesocarp, and fragments of the seed. To capture the night activity of the elephants and obtain individual identity and the time of faeces excretion, we used ten camera traps placed at various angles inside each of the three stables. In cases where camera traps could not be used to identify individual elephants, dung bolus diameter was recorded as a contingency measure in cases where individual elephant identification might not have been possible and infer the likely depositor based on known body size differences among elephants within each stable. Each paper envelope containing the retrieved seeds from each dung pile was clearly labelled and thus recorded: stable number, elephant identification, freshness score, approximate time and date of deposition depending on faeces freshness, and dung diameter from at least three boluses in each dung pile. The dung inspection and seed collection were stopped on day 5, when no more seeds were found.The germination experiment made use of seeds collected from dung during the feeding trial. An equal number of not ingested/control seeds (n=110) was randomly selected for use in the germination experiment. Seeds from each treatment (feeding and control) were planted using the ‘seed snail planting’ method. Four seeds are placed on top of a series of layers: one rectangular sheet of maize fibre, on top a moistened cardboard (100 cm x 60 cm), which is then covered by a layer of sand and commercial potting soil, followed by one more layer of moistened cardboard. The rectangle was rolled up like a sleeping mat, secured using aluminium wire, and arranged in an upright position inside buckets containing approximately 20 cm of water. Each seed was allocated a unique identification number.Statistical data analysesCumulative recovery curves were constructed to compare gut passage time for intact and broken seeds. To formally test if gut passage time differed between ingested and broken seeds, a non-parametric Wilcoxon rank-sum test was used (Figure 1).To examine whether gut passage time varied with elephant size, analyses were conducted using elephant-level seed recovery data. Only observations with a recorded gut passage time (days_since_feeding_trial) were retained, and records associated with unidentified elephant individuals were excluded. Elephant identity (name_ID_unknown) was treated as a categorical factor to allow visual differentiation among individuals. An exploratory visualisation was performed using a scatterplot of gut passage time (days) against elephant weight (in tonnes). Individual data points were plotted using jitter to reduce overlapping and to better visualise the distribution of observations across all elephants. Points were colour-coded by elephant identity to highlight within and among individual variation in gut passage time. To account for repeated observations for each elephant, the mean gut passage time was calculated for each identified elephant. A linear regression model was then fitted to assess the relationship between elephant size and mean gut passage time. Elephant body mass was included as a continuous predictor variable, and mean gut passage time (days) as the response variable. Model assumptions were evaluated visually through inspection of the fitted relationship (Figures 2 & 3)Because many retrieved seeds could not be assigned to an individual elephant, we used dung diameter as a proxy for elephant size. To test whether gut passage time of intact and broken seeds varied in relation to dung size, we examined the association between the number of days between seed ingestion and retrieval (days_since_feeding_trial) and the average dung bolus diameter (averagediameter). A simple linear regression model was fitted using ordinary least squares, with gut passage time as a continuous response variable and average dung diameter as the explanatory variable. Model assumptions were assessed through inspection of residuals, and model outputs were used to evaluate the significance of the relationship between dung diameter and gut passage time (Figure 4).To examine whether seed fate varies with elephant body size following gut passage, we conducted analyses at both the seed and the elephant levels. We included in the analysis only seeds for which elephant weight was known, excluding observations associated with unknown elephant weight. For each elephant, the number of intact seeds (See