Introduction
The savanna biome dominates southern Africa, and is characterised by continuous grassy understory, a discontinuous woody overstory (Archer et al., 2017; Bond, 2019), and, therefore, considered a largely open ecosystem (Pausas & Bond, 2020). However, various anthropogenic activities (Osborne et al., 2018) pose a threat to this biome. One pervasive threat is that of woody plant encroachment into savannas and grasslands, linked to climate change in particular (Criado, Myers-Smith, Bjorkman, Lehmann, & Stevens, 2020). Over the past two centuries (Archer et al., 2017) encroachment rates have varied between -0.131 to 1.275% per year in South Africa (O’Connor, Puttick, & Hoffman, 2014), and 0.25% per year across the rest of Africa (Stevens, Lehmann, Murphy, & Durigan, 2017).
Broad-scale studies predict that woody plant encroachment threatens biodiversity (Parr, Gray, & Bond, 2012), reducing the quality and quantity of the habitat, and jeopardizing both ecological functioning and services. In communal grazing lands and protected areas, bush encroachment also reduce production forage and increase ectoparasites (Archer et al., 2017). Overall, biodiversity is strongly affected by woody plant encroachment, e.g. in Grampians National Park, southern Australia, species richness of herbaceous growth forms was significantly reduced (Price & Morgan, 2008) while bird communities in the Kingdom of Eswatini showed significant shifts in structure (Sirami & Monadjem, 2012). This has real implications for conservation managers that aim to maintain diversity.
Wingless arthropods and those with low dispersal ability will be the worst affected by local habitat change (Blaum, Seymour, Rossmanith, Schwager, & Jeltsch, 2009). However, gradients of shrub cover vary in their impact, with grasshopper richness peaking at intermediate shrub cover, but not affecting butterfly richness (Koch, Edwards, Blanckenhorn, Walter, & Hofer, 2015). Hering et al. (2019) and Blaum et al. (2009) reported similar mixed effects on beetles, with some families increasing in abundance and richness at higher encroachment levels. Different metrics might also respond differently, e.g. shrub species in Australian woodlands were important drivers of arthropod community composition, but did not affect their abundance (Kwok & Eldridge, 2016). However, general consensus is that woody plant encroachment will have adverse effects on the ecological and ecosystem services (Eldridge et al., 2011) provided by arthropods, for example pollination (Kettenbach, Miller-Struttmann, Moffett, & Galen, 2017; Lara-Romero, Garcia, Morente-Lopez, & Iriondo, 2016), and dung decomposition (Blaum et al., 2009).
Ants (Hymenoptera: Formicidae), are among the most abundant terrestrial arthropods (Hölldobler & Wilson, 1990), and perform a myriad of ecological functions (Wills & Landis, 2018) and ecosystem services (Del Toro, Ribbons, & Pelini, 2012). Their response to encroachment has rarely been investigated and varies geographically from increased abundance in the southern Kalahari (Blaum et al., 2009) and increased richness in the south western USA (Bestelmeyer, 2005). However, Parr et al. (2012) found no real differences between ant richness and activity in open and closed habitats in Hluhluwe-Imfolozi Park, but observed that more species were restricted to closed habitats and that these restricted taxa were largely predatory (80% in the forest compared to the 7% in open habitats). Their results mainly focus on these rare taxa. Although they analyzed compositional differences, the distance-based approach used leads to confounding location and dispersion effects and has been shown to not reflect community composition (Warton, Wright, & Wang, 2012). Here, we use generalized linear models and their multivariate extensions to model ant species abundance (Warton et al., 2012). Ant functional roles is largely the result of dominant taxa in ant assemblages. Here, we complement their study by focusing on the community as a whole, and ask (1) how seasonality, soil and structural habitat complexity interact to affect ant diversity in an African savanna biome experiencing extensive woody thickening, (2) and identify species associated with these differences.