Study area
Field data were collected on farms around Clarens (28˚32'8''S; 28˚25'2''E) and Fouriesburg (28˚41'13''S; 28˚13'59''E) in the eastern Free State Province, South Africa. This area was selected for the study mainly due to its large P. angustifolia invasions in diverse habitats, accessible for research. The study area is typified by private farmland with some patches of natural grassland, rocky outcrops, and sandstone cliffs deeply incised by rivers (Mucina and Rutherford 2006). The area is surrounded by agro-industrial activities to the north, east and west; these include croplands, fallow agricultural fields that are no longer arable and some cattle ranching (Mucina and Rutherford 2006). Two field sites were chosen for observations. The first study field (Clarens) was used for camera traps and the second site (Fouriesburg) for timed observations.
Study species
Pyracantha angustifolia is native to south-western China (Cuizhi and Spongengberg 2003). Pyracantha angustifolia has been widely planted and has subsequently become naturalised in many countries including the high elevation, grassland interior of South Africa (Chari et al. 2020). In South Africa, P. angustifolia produces white flowers (5 mm in diameter) in October to December and orange-red or orange-yellow berries (5–8 mm across) in large numbers (Chari et al. 2020). In invaded sites, P. angustifolia can change habitat to dense monocultures (Daehler 1998). Five Pyracantha species have been introduced to South Africa as ornamental plants, with P. angustifolia being the most widespread among Pyracantha species (Moffet 2018). As indicated by Alien and Invasive Species Regulations of the National Environmental Management: Biodiversity Act, (South African Department of Environmental Affairs. 2014), all species are included as 1b Category invasive alien plants species (Department of Environmental Affairs 2014). Plants under category 1b must be controlled.
Field observations
We used camera-traps and opportunistic observations to determine which vertebrate species fed on P. angustifolia fruits during their fruiting season (July – September 2018). Remote camera-traps (Bushnell Trophy Cam HD Max-Colour LCD, Bushnell Outdoor Products, Overland Park, KS, USA) were used following methods by Zoeller et al. (2016) to identify vertebrate plant visitors, and the timing and frequency of their visits to feed on fruits. Three Bushnell remote camera-traps were mounted on tripods and aimed at fruiting branches of focal plants. Plants bigger in size (at least 4m high) were selected for observtions. Each plant was at least 30m away from the next plant. Each camera-trap was placed between 25 cm to one metre away from the. Camera-trap storage capacity and battery strength were checked at least once a week and data downloaded. Camera-traps were put in the field for ten weeks (11 July – 23 August 2018 and 03–24 October 2018). The sum of all hours that all cameras were active totalled 1 826 hours. The remote camera-traps were triggered by heat or motion and would take photographs and videos. When triggered, the camera-traps filmed the visitors’ activities for 60 seconds, and could be re-triggered after a 10 second interval. From the videos, the potential seed dispersers were identified using bird field guide books (Hockey et al. 2005; Newman 2013). The number, species and behaviour of visitors, and duration of foraging bouts were noted.
We conducted direct observations to complement the data captured by the camera-traps. Opportunistic observations were conducted in Clarens. Frugivore visitors to P. angustifolia were identified using bird field guide books (Hockey et al. 2005; Newman 2013) and their behaviour observed. Only birds observed feeding on P. angustifolia fruits were noted in our opportunistic observations.
Timed observations of avian frugivores were conducted at Wynford Holiday Farm (28˚41'13''S; 28˚13'59''E), located between Fouriesburg town and the Lesotho border. This former grassland area has been transformed into a dense monoculture of P. angustifolia (L. Adams, pers. obs.). Observation periods spanned six hours a day, consisting of a three hour period of peak bird activity after sunrise (07:00–10:00) and a three hour period of peak activity before sunset (14:00–17:00) over three days (total of 54 h) in June 2019. We chose three mature fruiting shrubs for observations, observed birds using binoculars and, if possible, took photographs of visiting birds. For each observation, we noted foraging bird arrival and departure time, and behaviour (feeding or resting).
Bird feeding trials
Fresh, ripe P. angustifolia fruits were collected from 20 shrubs from the Clarens area (28˚32'69''S; 28˚25'20''E) during the fruiting season (August 2019). Branches with fruits were collected to keep them as fresh as possible. Fruits were removed from branches by hand and placed in plastic bags at the Animal House at the University of KwaZulu-Natal (UKZN), Pietermaritzburg campus, KwaZulu-Natal Province, South Africa. Fruits from all shrubs were pooled, in order to standardise any differences associated with local seed origin as well as incorporating genetic variability (Tecco et al. 2016). After collection, fruits were placed in a refrigerator, and the feeding trials started a day after collection.
Bird species chosen to be used for feeding and germination trials were informed by the field observations. Bird species used in the germination study included five Dark-capped Bulbuls (Pycnonotus tricolor), two Purple-crested Turacos (Gallirex porphyreolophus), three Speckled Mousebirds (Colius striatus), five Red-winged Starlings (Onychognathus morio) and nine Cape White-eyes (Zosterops virens). All bird species had been observed feeding on P. angustifolia fruits in the field except for Purple-crested Turacos. Purple-crested Turacos and Red-winged Starlings were regarded as relatively large birds as their mean body mass was 285 and 140 g respectively (Newman 2013). In comparison, Speckled Mousebirds, Dark-capped Bulbuls and Cape White-eyes were regarded as relatively small birds weighing on average 55, 39 and nine grams respectively (Newman 2013). Purple-crested Turacos do not naturally occur in the study area, but we used them as a surrogate species for other larger bird species that consume the fruits in the study area; for example, the African Olive Pigeon (Columba arquatrix) (L. Adams, pers. obs).
Captive birds used in our study were kept in outside aviaries (1 x 2.12 x 2.66 m) at the UKZN Animal house. They were then moved and acclimatised to experimental indoor cages in a constant environment room with 12L: 12D photoperiod at 25 ± 1˚C (Thabethe et al. 2015) for five days before the commencement of feeding trials. They were fed a maintenance diet daily, including apples, pears, bananas, and oranges, together with a bird protein supplement of AviPlus Softbill/Mynah pellets (Aviproducts, Durban, RSA).
Fruits were fed to birds, and the germination success of defecated seeds investigated. Feeding trials were conducted for three days after the acclimatisation period. Birds were given their supplementary food incorporated with five P. angustifolia fruits. Supplementary food was removed from the cages at 17h00 as the birds do not feed at night and would be hungry in the following morning for the feeding trials (Dlamini et al. 2018). Feeding trials began at 08h00 where birds were provided with P. angustifolia fruits to feed on. Water was provided to birds ad libitum.
Seed retention time
To inform on potential seed dispersal distances, seed retention data were collected to predict the effect of seed dispersal distances. The seed retention time was determined for each bird species to investigate how long seeds took to pass through the birds’ gut. This measure was to help predict the spatial implications of seed dispersed by birds. The seed retention time was recorded as the time from when the bird consumed the first fruit/seed until the first seed was defecated by the bird.
Germination trials
Different germination treatments were used in order to compare the germination success and rate. Treatments included seeds ingested by birds, manually depulped seeds where the fruit wall was removed by hand, and intact fruits. The birds were only observed during feeding until sufficient numbers of seeds were gathered (at least 50) for planting (typically 4h). We placed newspaper under the bird cages to trap all seeds excreted by birds. All seeds were collected and kept in paper bags before planting. Seeds collected from bird excreta were planted in potting soil in seedling trays (265 × 180 × 75 mm) and kept in a greenhouse at 26 ˚C maximum temperature (Jordaan et al. 2011). The sown seeds were watered daily and were considered to have germinated when the radicle appeared above the soil (Dlamini et al. 2018). After being counted, such seedlings were removed from the seedling tray to avoid recounts (Dlamini et al. 2018).
The treatment using depulped seeds emulated a condition where the fruit wall was removed by other factors but not birds’ ingestion. This treatment excluded the digestive juices or abrasion in the frugivores' gut as factors that might trigger germination. A total number of 200 seeds were planted from 40 manually depulped fruits as per above.
Un-ingested intact fruits were sown in the same conditions as the other two treatments to serve as a standard of comparison (control) for the experiment. This treatment emulated the natural environment where seeds fall directly to the soil without being eaten by animals, buried under the soil or litter while intact. One hundred fruits (n = 100) were planted in four separate planting trays, totalling approximately 500 seeds (average of 5 seeds per fruit). Seedling emergence was recorded until the 58th day for all three treatments.
Data analyses
We used GraphPad Prism 5 statistical software (Motulsky 1999) to analyse data for the study. We calculated mean cumulative seedling percentages for ingested, manually depulped and whole fruits. Time taken for seedlings to emerge in each treatment were compared using analysis of variance (ANOVA) (Thabethe et al. 2015). Mean seedling emergence times for seeds ingested by different bird species, manually depulped, and whole fruits (control) were compared using post-hoc Tukey tests for multiple comparisons. Seedling emergence time referred to the time it takes for a single seedling to emerge on top of the soil. The significance level for this analysis was set at α = 0.05. We conducted a two-tailed Chi-square test at 95% confidence interval to compare the proportion of seeds that germinated from each treatment. Data normality was tested using a D’Agostino and Pearson omnibus normality test (D’agostino et al 1990).