Disturbances in drylands: Interactions among herbivory, drought, and termite activity in savanna plant communities

1 INTRODUCTION

Disturbance—any biotic or abiotic force that generates deviations from prevailing local background conditions (Graham et al., 2021)—has long been recognised as a fundamental driver of population dynamics, species distributions, and community structure (Pickett & White, 1985; Sousa, 1984). Although the impacts of many types of disturbance have been studied in isolation, the individual and net impacts of multiple interacting disturbances are difficult to disentangle (Turner, 2010). Whether such interactions are additive or non-additive—and in the latter case, whether super-additive or sub-additive (Crain et al., 2008; Piggott et al., 2015)—is rarely established (Porensky & Young, 2013). The potential for non-additive effects and emergent properties in complex systems has important implications because it determines whether or not the consequences of multiple disturbance agents can be predicted from the study of each individually (Buma, 2015; Burton et al., 2020).

The need to understand the impacts and interactions of multiple disturbances is urgent given accelerating change in climate, land use and community structure. Beyond the steadily shifting climatic baseline, models predict increases in the frequency and/or intensity of extreme-weather events (IPCC, 2022), which have severe and long-lasting ecological consequences (Anderegg et al., 2015; Walker et al., 2023). Similarly, the millennia-long decline of large-bodied mammals is currently being punctuated by a spasm of population extirpations, which sharply transform local disturbance regimes (Pringle et al., 2023; Ripple et al., 2015; Smith et al., 2018). The impacts of shifting disturbance regimes may be especially acute in drylands, which cover over 40% of global land area and are often sensitive to even small changes in bottom-up (precipitation) or top-down (herbivory) forces (Maestre et al., 2016, 2022).

We investigated the effects of three dominant drivers of vegetation dynamics in tropical semi-arid grasslands: large-mammal herbivory, rainfall (drought), and soil-nesting termites. Of the possible interactions among these drivers, herbivory × rainfall interactions are the most extensively studied (Carmona et al., 2012; Ebel et al., 2022; Fuhlendorf & Smeins, 1997; Hartvigsen, 2000; Milchunas et al., 1989) and are often non-additive, such that the effects of one are amplified by the other (Augustine & McNaughton, 2006; Gao et al., 2009; Koerner & Collins, 2014; Loeser et al., 2007; Porensky et al., 2013; Riginos et al., 2018). However, the extent to which termite activity modifies the effects of herbivory and/or rainfall remains unclear.

Termites are predominant decomposers in ecosystems world-wide (Zanne et al., 2022) and are ubiquitous in African savannas (~100 kg ha⁻¹, similar to mammalian herbivores; Moe et al., 2009). The centralised nests (termitaria or ‘mounds’) created by fungus-farming termites (Macrotermitinae) are particularly influential in Paleotropical savannas and affect vegetation at local to landscape scales owing to alterations of soil texture, nutrients and moisture (Pringle et al., 2010; Pringle & Tarnita, 2017; Tarnita et al., 2017). While termitaria typically increase localised productivity, their effects on understorey cover and diversity are variable (Davies et al., 2014; Muvengwi et al., 2017; Muvengwi & Witkowski, 2020; Okullo & Moe, 2012a, 2012b). Termitaria × rainfall and termitaria × herbivory interactions have occasionally been documented. For example, termitaria can enhance grass cover more in drier habitats (Davies et al., 2014), maintain grass cover primarily during the wet season (Okullo & Moe, 2012b), and attract large herbivores (Davies et al., 2016; Odadi et al., 2018).

Theoretical models have predicted that termitaria should enhance the robustness of dryland vegetation to drought by enabling plants to withstand (resistance) and/or recover from (resilience) water limitation (Bonachela et al., 2015; Castillo Vardaro et al., 2021; Tarnita et al., 2017). While some empirical evidence is consistent with this prediction (Ashton et al., 2019; Guirado et al., 2023), it is difficult to test directly. In addition, few studies have explored the role of herbivory in mediating termite (Okullo & Moe, 2012a, 2012b; Trisos et al., 2021). For example, productive conditions on termitaria might help plants tolerate herbivory; however, given that the capacity for compensatory regrowth is constrained by water limitation and that herbivory at termitaria often intensifies during dry periods (Daskin et al., 2023; Davies et al., 2016), any buffering effects of termitaria may be negated or outweighed by interactions with bottom-up (drought) and top-down (herbivory) stressors (Trisos et al., 2021). Three-way interactions among rainfall, herbivory, and termitaria have not, to our knowledge, been experimentally evaluated and might provide insights that are overlooked by studies of individual or pairwise effects (Kercher & Zedler, 2004; Koerner & Collins, 2014).

Plant responses to these three drivers (herbivores, rainfall, and termitaria) are likely to depend on plant functional traits, such as life form and life history. For example, herbivory often benefits annuals relative to perennials and alters the balance of forbs and grasses (Anderson et al., 2007; Loeser et al., 2007; Pakeman, 2004; Pérez-Camacho et al., 2012), and annual grasslands may be more sensitive to interannual variation, including droughts, than perennial grasslands (Ruppert et al., 2015; Werner et al., 2024). However, such studies often analyse aggregate responses of functional groups, which are driven by their dominant species, and thus do not necessarily enable reliable inferences about the responses of species of the same functional type in other similar ecosystems.

The variable interactions among herbivory, termitaria and drought documented in previous studies may stem in part from their short duration. This is particularly pertinent for episodic disturbances, such as extreme weather, because studies encompassing just one event cannot robustly test interactions. Moreover, communities dominated by slow-growing perennial plants can take years for experimental effects to manifest fully (Porensky et al., 2017; Riginos et al., 2018). The handful of active multi-decade experiments is therefore inordinately valuable for untangling multi-disturbance interactions, especially those that are intermittent and unpredictable (Gaiser et al., 2020).

Since 1995, the Kenya Long-term Exclosure Experiment (KLEE) has manipulated large-mammal herbivory in replicated 40,000-m² plots of semi-arid African savanna (two to three orders of magnitude larger than the typical exclosure experiment; Pringle et al., 2023). Using 15 years of data from KLEE and 2 years of data from an adjacent small-scale resource-addition experiment with crossed herbivore exclusion, water-addition, and fertilisation treatments, we applied hierarchical multispecies models to test (a) the individual and interactive effects of herbivory, rainfall, and termites on understorey vegetation cover and species richness, and (b) how these responses were mediated by plant life-form (graminoid/forb) and life history (annual/perennial). We hypothesised a three-way interaction among herbivory, drought, and termitaria, whereby the more-than-additive suppression of understorey plants by herbivory and drought would be dampened on termitaria.