Invasive trees and shrubs represent a major threat to natural resources in South Africa (Richardson & van Wilgen, 2004; Gaertner et al., 2009), and their currently increasing numbers (Richardson & Rejmánek, 2011), suggest that their threat might intensify in future. Of the approximately 70 Acacia tree species that have been introduced into South Africa, predominantly during the 19th Century, at least 13 are currently recognised as invasive (Henderson, 2007; Richardson et al., 2011). Of these, Acacia cyclops commonly known as Rooikrans, was introduced into South Africa in 1835, primarily for stabilisation of sand dunes in the south-western Cape region (Shaughnessy, 1980). It has since become naturalised and has invaded coastal regions across the southern Cape (Henderson, 2007). The proliferation of A. cyclops in South Africa is ascribed to it having been widely propagated as a source of fuel wood for domestic and small-scale commercial purposes (Richardson et al., 2011), and to its prolific seed production (Milton and Hall, 1981; Impson et al., 2011) which increases chances of dispersal by vertebrates. Each seed has a nutritional and orange-reddish aril which attracts vertebrates, especially birds that consume and pass the indigestible seeds and thereby enhance the population dynamics of the plant (Glyphis et al., 1981; Underhill and Hofmeyr, 2007; Mokotjomela and Hoffmann, 2013). Indeed, recent study has shown that several bird species including selected ubiquitous dove species improve germination rates of ingested seeds (Mokotjomela et al., 2015). The seeds are also buried underground in accumulations by ants (Holmes, 1990) and possibly seed-hoarding rodents (Midgley and Anderson, 2005), and thus reducing impacts of predation and incineration of seeds during periodic fires. Holmes (1990) reported that A. cyclops seed banks tend to increase with vegetation cover and thus suggesting dense foliage and tree canopy limit predators’ access for seeds.

Biological control agent species including Melanterius servulus (a seed-feeding weevil) and Dasineura dielsi (a flower-galling midge) have been released on A. cyclops in South Africa in 1991 and 2002, respectively. Both agents were imported from Australia (the native range of A. cyclops), and they are prolific, causing high levels of seed damage in A. cyclops which in turn is expected to limit the invasiveness of the species (Impson et al., 2011; Moran et al., 2013). For instance, M. servulus inflicts seed damaged of up to 97.5%, with an overall average reduction of seed production of 56% between 1998 and 2009 in the South-western Cape (Impson et al., 2011). An earlier study at Koeberg Nature Reserve during the 2012 seeding season showed that invertebrate (e.g. ants) removal of seeds was much lower under dense stands of A. cyclops (Mokotjomela and Hoffmann, 2013) than in the 1980s (Holmes, 1990). However, it is not known how seed reduction imparted by the biological control agents has affected patterns of granivory by other organisms that potentially influence recruitment processes in A. cyclops.

Foraging patterns in vertebrates, especially rodents and other organisms, are influenced by degree of food availability (Bartness and Albers, 2000) and environmental temperatures (Bartness and Albers, 2000; Feldhamer, 2007; Orrock and Danielson, 2009). Consistently, the frugivorous birds in the Mediterranean climate region in South Africa foraged on exotic species with abundant fruit resources (Mokotjomela et al., 2013). Also, foraging patterns of the murid rodent, Ice rat (Otomys sloggetti) in the Drakensberg Alps was restricted to warm parts of the day (Hinze, 2005) while the desert rodents in southern Africa choose to forage during cooler times to avoid thermal stress (Feldhamer, 2007). Because vegetation cover may change the habitat quality for seed predators by modifying, for example, the microclimate (e.g. light, temperature) and soil characteristics (e.g. plant litter), and the risk of predation (Manson and Stiles, 1998), granivory rates tend to be higher under cover than open areas (Meiss et al., 2010). In addition, seed removal and dispersal by ants tend to be more common in hot and semi-arid areas in South Africa than in mesic areas (Bond and Slingsby, 1983). Feeding times of the day on seeds (i.e. morning, midday and afternoons) have important influence on potential dispersal of the ingested seeds in birds (Bibby et al., 2000; Kays et al., 2011), with the seeds of the Panamanian nutmeg tree (Virola nobilis), for example, consumed in the mornings by birds standing a greater chance of dispersal than those consumed later in the day (Kays et al., 2011). Furthermore, feeding peak time in the afternoon have disadvantage since the ingested seeds are then deposited at the roosting sites where seed mortalities are high due to predators and competition between seedlings (Westcott et al., 2005; Kays et al., 2011; Mokotjomela et al., 2013).

Despite the knowledge that birds are important seed dispersers of the highly invasive A. cyclops in South Africa (Glyphis et al., 1981; Underhill and Hofmeyr, 2007; Mokotjomela and Hoffmann, 2013; Mokotjomela et al., 2015), there is a limited knowledge on the temporal patterns of granivory in A. cyclops stands that have seen seed reduction due to biological control agents, and how habitat characteristics such as vegetation cover may influence the patterns of granivory. Granivory of seeds of A. cyclops has important implications on seed banks dynamics which reportedly account for its invasion success, since a larger proportion of the seed crop falls on the ground when pods dry up (Holmes, 1990). The major aim of the present study was to determine how granivory rates in A. cyclops are affected by vegetation cover during different phases of seeding period associated with variable abundance of seeds, and different times of day, which have implications for seed dispersal probabilities. This study tested predictions that rodents’ granivory might be higher in seed cages located under dense tree canopy cover, where predation risk is relatively lower than in those seed cages located in open and low vegetation cove, and that invertebrates’ seed removal might predominate due to semi-arid to arid conditions of the study site.

Seed consumption by vertebrates has important implications on seed fate and recruitment in plants (Nathan and Muller-Landau, 2000), and thus understanding granivory rates might be important for conservation management where invasive alien plants threaten indigenous species. This study investigated how vegetation cover affected granivory patterns and rates over different phases of the seeding season (e.g. bi-monthly) and times of day in an invasive shrub A. cyclops stands under biological control in the South-western Cape, South Africa.

Contrary to the study prediction, vegetation cover did not influence seed removal by vertebrates including rodents. An equivalent seed removal by rodents both under and outside tree canopy might reflect a limited seed availability, associated in part with biological control effect (Impson et al., 2011), and partly due to competition for seeds resource by other organisms whose foraging activity is restricted in the forest either due to fear of predation or extreme temperatures during the daytime. Thus, rodents seeking seeds extend their foraging range beyond the forest canopy and thus confound the reported effects vegetation cover on their foraging patterns (Meiss et al., 2010). The relatively low seed removal by rodents during the early phases of the seeding period in this study is a likely result of high abundance of A. cyclops seeds, of which Holmes (1990) reported to be causing predator satiation. It is therefore logical that the peak seed removal by rodents occurred during late phases of the seeding period when the number of available seeds was declining.

In spite of the biological control agents that have substantially reduced reproductive output in A. cyclops stands in South Africa, seed removal and dispersal still occur (Mokotjomela et al., 2015). This report was confirmed by the finding that rodents (e.g. striped mice Rhabdomys pumilio), which are key seed consumers of A. cyclops (Holmes, 1990; Mokotjomela and Hoffmann, 2013), removed fewer seeds than were removed from the experimental control seed cages. Indeed, additional organisms including birds and large mammals that are known to be effective seed dispersal agents were documented feeding on A. cyclops post-dispersal seeds at Koeberg Nature Reserve (Mokotjomela and Hoffmann, 2013). Nevertheless, birds and mammals removed relatively fewer seeds (10%) compared to a previous report showing 15% removal (Mokotjomela and Hoffmann, 2013). A possible explanation is high competition for scarce seed resource in combination with low abundance of granivorous organisms imparted by the heavily fragmented local environment at Langebaanweg. This finding also suggests that a limited number of seeds might be dispersed by birds from A. cyclops stands having seed production reduced by biological control agents (Mokotjomela et al., 2015). Removal of seeds during the cooler afternoon times might limit long distance dispersal, and increase mortality, since Kays et al. (2011) reported that birds which are effective dispersal agents, tend to be most active in the mornings.

That invertebrates removed the fewest seeds among the treatments, and that this happened in seed cages located in open/sparse canopy cover, contradicts the second study prediction that invertebrates’ seed removal especially by ants would be predominant due to prevalent aridity of the Langebaanweg (Bond and Slingsby, 1983), and a previous study (e.g. Holmes, 1990) showing that ants remove more seeds under dense vegetation cover. This finding could be explained by interference competition by multiple organisms taking shade cover under the trees during extremely hot daytimes. Trampling of the seed aril which is a target resource for ants, by larger organisms (Mokotjomela and Hoffmann, 2013), might render the aril unfavourable to ants. Also, the seed aril deteriorates rapidly due to microbial effect in relatively moist ground conditions under dense canopy cover (Meiss et al., 2010; Fricke et al., 2013), thereby losing chemical compounds which Midgley and Bond (1995) suggested play an important role in attracting ants and thus mediating their seed removal activity in Acacia species. To confirm the importance of fresh nutritious A. cyclops’ seed aril to ants and other invertebrates, greater numbers of seeds were removed during the beginning of the seeding phase (i.e. after pods’ opening) in this study.

In conclusion, the study has shown that uniform granivory patterns of rodents under closed and open canopy covers might be an indication of reduced seed production. This, however, contradicts reported trends showing importance of vegetation cover as refugia for foraging rodents (Andrzejewski, 2002; Meiss et al., 2010). Because a large portion of A. cyclops seed crop falls on the ground, where dispersal and ground seed bank recharging occur (Holmes, 1990), the observed high rodents’ seed predation may be a limiting factor to both processes, and thus complementing the biological control agents. Seeds of A. cyclops contribute more than 50% of the daily stripped mice's diet in the South-western Cape (Holmes, 1990). Alternatively, invertebrates continue to constantly remove and bury seeds, since different studies (e.g. Holmes, 1990; Mokotjomela and Hoffmann, 2013 and this study) have reported equivalent seed removal rates of approximately 13–16% at Table Mountain and Koeberg Nature Reserve and Langebaanweg respectively. Taken together, these studies suggest that ants’ seed dispersal at least for A. cyclops might not be significantly influenced by environmental conditions, including moisture and temperature. Although these results are site-specific, they provide preliminary highlights on success of biological control method of controlling invasive alien plants in South Africa. Indeed, Richardson et al. (2011) predicted that Australian acacias have potential to invade about a third to global terrestrial environment and threaten biodiversity of which suggest they require stringent management and control approaches such biological control agents to pro-actively suppress their invasions where they are being used for commercial purposes.

The author declares no conflicts of interest.