Mixed-species foraging flocks (MSFFs) of birds can be defined as aggregations of more than two species that actively initiate and continue their association (Harrison & Whitehouse 2011) while foraging in an area or habitat (Greig-Smith 1978). Despite the global prevalence of MSFFs among terrestrial bird communities (Harrison & Whitehouse 2011; Morse 1970) MSFFs have rarely been characterised in southern Africa. The presence and role of each species in a flock reflects its ecological niche as well as the various factors leading to flock formation (Morse 1970).

In temperate areas, MSSFs tend to form more readily in non-breeding seasons (Morse 1970) and differ from other aggregations of birds in their cohesion, rather than being a gathering of different species drawn to a particular resource (such as a fruiting tree); these flocks remain together as they progress through the habitat (Goodale & Beauchamp 2010). Only those individuals that lead the flock or make a concerted effort to follow and remain in the flock should be considered as MSFF members (Greig-Smith 1978). MSFFs are typically led by a particular species, a role typically fulfilled by a so-called ‘nucleus species’. A nucleus species is usually a species with high flocking propensity, encountered in a high percentage of MSFFs, often occurs in intra-specific flocks, and usually displays conspicuous behaviour such as vocalising or making conspicuous movements (Hutto 1994). The individuals of a nucleus species are wholly or partially responsible for the formation and continued cohesion of the MSFF (Goodale & Beauchamp 2010; Greig-Smith 1978; Morse 1970) with other species in the MSFF organising their various activities around the nucleus species. Although various feeding guilds are present, MSFFs are usually dominated by insectivorous species (Croxall 1976; Greig-Smth 1978).

In some habitats multiple types of MSFFs may be distinguished, based on differences in foraging height (Zou et al. 2011) or the presence of particular species or guilds (King & Rappole 2001). Although some authors have suggested that the regular occurrence of a given species in MSFFs is directly related to the abundance of the species (Morse 1978; Zou et al. 2011), this is not always the case in savanna ecosystems (Greig-Smith 1978). Furthermore, some species may be recorded in MSFFs only when the flock happens to move through the territory of that species (Greig-Smith 1978). The mean number of species per MSFF ranges from 3.8 in forested areas in China (Zou et al. 2011) to 7.7 and 9.9 in the former eastern provinces and Barotse of Northern Rhodesia respectively (Winterbottom 1943). Furthermore, although differences between MSFFs observed in different vegetation types within one geographical area may be minimal (Croxall 1976; Winterbottom 1943), MSFFs in adjacent Acacia and mixed woodland habitats in northern South Africa do show significant differences in composition and mean number of species (Thomson & Ferguson 2007).

To date, MSFFs occurring in any of the various savanna habitats of the Kruger National Park (KNP), South Africa have not been investigated despite the major role that birds play in savanna ecosystems and the maintenance of savanna heterogeneity (Kemp et al. 2003). The main aim of this study was to investigate the composition of MSFFs in a southern section of KNP.

During the study period, 62 MSFFs were encountered, equally divided between the two habitat types (i.e. 31 MSFFs in Acacia and 31 MSFFs in Combretum). In total 1251 individuals of 74 different species were observed in these MSFFs. Of these 74 species, 29 species in Acacia and 24 species in Combretum were regarded as regular flocking species (species that occurred in > 10% of flocks; King & Rappole 2001). Only these species were used for further analysis (Table 1). Of the 32 (43%) species that were regarded as regular flocking species, 21 species (28%) were recorded in both the Acacia and the Combretum, 8 (11%) in the Acacia only, and 3 (4%) in the Combretum only.

Across flocks the mean number of individuals per species was 9.3 (± 4.5 s.d.) and 7.6 (± 5.6 s.d.) in the Acacia and Combretum habitat types respectively. Results from the Wilcoxon rank sum test indicated that MSFFs in Acacia had significantly (P < 0.0001, n = 31, Z = 4.2) more individuals per MSFFs (mean: 21.5 ± 12.6 s.d.) than those in Combretum (mean: 10.7 ± 5.2 s.d.). A significant difference (T = 3.91; P = 0.0003) in the number of species per MSFF between the Acacia and Combretum was also found, with the Acacia having a mean number of species per MSFF of 8.7 (± 3.5 s.d.) and the Combretum 5.9 (± 1.7 s.d.).

The frequency of occurrence of a recorded species in the MSFFs often varied according to habitat type. The three most frequently occurring species in Acacia MSFFs were Fork-tailed Drongo (Dicrurus adsimilis), Southern Black Tit (Melaniparus viger), and Chinspot Batis (Batis molitor); whereas in Combretum they were Fork-tailed Drongo, Rattling Cisticola (Cisticola chiniana), and Chinspot Batis (Table 1).

There were 26 significant correlations among species pairs in the Acacia habitat of which 20 were positive (Table 2). Of these, the strongest associations are between Southern Yellow-billed hornbills (Tockus leucomelas), Fiscal Flycatchers (Melaenornis silens), and Southern Red-billed Hornbills (Tockus rufirostris); Black-backed Puffbacks (Dryoscopus cubla) and Tawny-flanked Prinias (Prinia subflava); Southern Grey-headed Sparrows (Passer diffuses) with Fiscal Flycatchers; and Southern Masked Weavers (Ploceus velatus), with Arrow-marked Babblers (Turdoides jardineii).

In the Combretum habitat there were nine significant correlations among species pairs of which five were positive (Table 3). The following positive correlations were found in the Combretum habitat: Long-billed Crombecs (Sylvietta rufescens) with Rattling Cisticola, Emerald-spotted Wood Doves (Turtur chalcospilos) and Blue Waxbills (Uraeginthus angolensis) with Arrow-marked Babblers, Southern Grey-headed Sparrows with Yellow-fronted Canaries (Crithagra mozambica), Green Wood Hoopoes (Phoeniculus purpureus) with Black-backed Puffbacks. Negative correlations in the Combretum habitat included those between Chinspot Batis’ and Arrow-marked Babblers; Red-billed Firefinches (Lagonosticta senegala) with Arrow-marked Babblers; Greater Blue-eared Starlings (Lamprotornis chalybaeus) with Fork-tailed drongos; and Cardinal Woodpeckers (Dendropicos fuscescens) with Rattling Cisticolas.

The MSFFs observed during this study had mean MSFF sizes similar to the mean MSFF size (16.9–19.3) recorded in tropical forests in Myanmar, Southeast Asia (Croxall 1976; King & Rappole 2001) and west African savannas (23.2) (Greig-Smith 1978). The mean number of species per MSFF was comparable to MSFFs studied in other African savannas (Winterbottom 1943). Acacia supported larger flocks and a greater number of species per flock when compared to the Combretum. MSFFs in adjacent Acacia and mixed woodland habitats in northern South Africa also showed significant differences in composition and mean number of species (Thomson & Ferguson 2007). Acacia also exhibited a higher mean number of species per MSFF, possibly because of having greater structural complexity than Combretum. Species diversity typically increases with structural complexity of habitat (Skwono & Bond 2003).

Testing this hypothesis is however beyond the scope of this study, and reasons for these differences in MSFF sizes and species numbers between habitats would make for interesting future study. Having said this, the majority of species recorded in MSFFs in this study were recorded in both habitat types surveyed. In some cases, inconspicuous MSFF participants such as Stierling’s Wren-Warbler (Calamonastes stierlingi) may have been under-recorded because of the dense nature of the habitats in which they occurred. This would lead to underrepresentation of these species in the results. Although such species could have been actively revealed by flushing them, this would have led to MSFF dispersal and was thus deemed unsuitable.

In both vegetation types, Fork-tailed drongos occurred in 71% of MSFFs, a similar finding to that of many other studies where drongos were recorded as common MSFF participants (Croxall 1976; Greig-Smith 1978; Swynnerton 1915; Winterbottom 1943; Zou et al. 2011). In the MSFFs recorded in Nyslvley Nature Reserve, South Africa drongos however occurred in up to 92% of MSFFs (Thompson & Ferguson 2007). Drongos have been recorded as a nucleus species of MSFFs in various studies (Cordeiro et al. 2015; Goodale & Kotagama 2006; Nefdt 1989), and because of their prevalence in MSFFs in this study, drongos are therefore a very likely nucleus species in MSFF of the southern KNP. Other studies have however found that despite drongos occurring in MSFF they do not always take on the role of a nucleus species within a MSFF (Greig-Smith 1978). Of the other frequently encountered species, the Chinspot Batis and Southern Black Tit are notable because both batises (Batis spp.) and tits (Parus spp.) are frequent participants of MSFFs in savannas and forests elsewhere in Africa (Greig-Smith 1978; Nefdt 1989; Thomson & Ferguson 2007; Winterbottom 1943). In both Acacia and Combretum there was a clear tendency of certain species to be present in the majority of MSFFs. The Chinspot Batis, Southern Black Tit, and Fork-tailed Drongo are frequent participants in both habitat types. The regular occurrence of pairs of Fiscal Flycatchers in MSFFs in the Acacia habitat is noteworthy as this species is an uncommon winter visitor to the KNP (Newman 1991) and this vegetation type probably represents a favoured habitat for these flycatchers when they occur temporarily in this region. The season during which this study was conducted excluded all summer migrants and a similar investigation in the wet season may therefore reveal a different MSFF composition.

The non-random patterns of positive associations among observed MSFFs in this study support findings by King and Rappole (2001) that members of some species tend to associate with other species. King and Rappole (2001) suggested that the reasons for these positive associations between species in MSFFs was to facilitate more efficient foraging by flushing prey from the vegetation through which they move. The reasons for the positive association between MSFF members in this study are likely to be similar, but need further investigation to verify this.

This study is the first of its kind to be undertaken in the KNP and forms the baseline for ongoing work investigating the composition of MSFFs occurring in the broader landscapes of this region. The analysis of MSFFs could form a valuable component of the biodiversity monitoring currently being undertaken in the KNP, and any future change in the composition and behaviour of MSFFs may be an early indicator of threats to the biodiversity of this region.