PIP3 radio tags were fixed onto the middle tail rectrix of three adult Blue Swallows (one male and two females) and radio tracked for 5, 8 and 9 days, respectively. Radio tags consisted of an Ag317 battery, which powered the tags for up to 14 days, and a whip antenna measuring 116 mm in length and weighing between 340 mg and 360 mg (within the accepted threshold of 3% – 5% body weight [Cochran 1980; Kenward 2001]).

Two independently located base stations for direction finding (Kenward 2001) were set up for each site. Base station sites were chosen primarily for their proximity to the nest site and visibility over the surrounding landscape. A 360^o plastic compass was fixed to the foot of the antenna (a flexible three-element Vagi antenna [Lintec Antennas Ltd., West Sussex, United Kingdom] weighing 500 g with a bandwidth of approximately 2 MHz, a beamwidth of 80° and a gain of 6 dB stand). The base stations were in line of sight of each other to enable the correct base line zeroing (i.e. the 360^o at the first base station faced the second base station and vice versa). A metal peg was driven into the ground at the 360^o mark so that if inclement weather prevented visibility between the base stations, an accurate zero degree baseline could be re-established. The readings were made by inserting a needle through the base of the Vagi antenna stand, which pointed to the direction, in degrees, where the Vagi antenna was aimed.

The maximum foraging home range for breeding Blue Swallows was assumed to be the maximum distance obtained using radio tracking for a Blue Swallow tied to its active nest site. A circular area of 525 ha was calculated as the potential feeding habitat around an active nest site, using 1296 m (the furthest recorded distance) as the radius.

The nest locality data were obtained using a Global Positioning System (GPS), set up according to WGS84. These data were projected, using the Projector extension in ArcView, into Transverse Mercator in WGS84 and overlaid onto digital orthophotos. Land cover types were digitised on-screen and then cleaned using the ArcView extension Edit Tools version 3.3. The digitised land use data were then overlaid with the positional data for each radio tracked Blue Swallow. A Digital Terrain Model was used to determine the habitat in line of sight from each base station which was then combined to spatially delineate a catchment area surrounding each active nest that determined the potential feeding habitat that could be radio tracked.

The land cover on and adjacent to the study site was digitised as polygons, and classed into a land cover type based upon the homogeneity of the physical attributes of the type of land cover. A simplified label was used to describe the classification of the land cover (Table 1).

From field observations, it appeared as if Blue Swallows tended to increase their activity along the boundaries between the grassland and wetland habitats. To corroborate this observation, the distances of the points to an ecotone of a habitat were determined for the 940 radio telemetry points collected from the birds. The distances of 940 randomly generated points were then compared with the distances of the 940 observed points. A non-parametric Chi-square goodness of fit test (Χ²) was used to statistically compare the radio telemetry point data set and the number of expected locality points occurring within each of the habitat types.

Ethical clearance was granted by Ezemvelo KwaZulu-Natal Wildlife and the Endangered Wildlife Trust Blue Swallow Working Group (EWT-BSWG).

A total of 940 positional data points were gathered over 22 days for the three nest sites. The Florida bird had the largest foraging home range and both the Florida and Tafeni birds’ foraging home ranges overlapped with that of the Diptank bird (see Figure 2). The Diptank and Florida birds foraged further from the nest (1030 m and 1296 m, respectively) than the Tafeni bird (524 m). For spatial composition of the foraging home ranges for the three nest site areas, see Table 2. A total of 58% of the total observed locations occurred within 20 m of all habitat boundaries and 69% within 30 m (Figure 3a). In contrast to these findings, the randomly generated expected frequencies delivered 31% and 43%, respectively (Figure 3b). The Chi-square test produced highly significant findings (p < 0.001) between the observed and expected frequencies.

Grassland was the largest land cover type at the Diptank and Tafeni nest sites (104 ha and 59 ha, respectively, Table 3), while sugar cane fields (39 ha) and grassland (33 ha) were the largest land cover types at the Florida nest site. Of the 940 viewshed data points obtained, grassland habitat was the highest habitat type represented across all three nesting sites with a total of 708 points (75% in total and 75%, 69% and 80% for Diptank, Florida and Tafeni nest sites, respectively). As a land cover, commercial timber plantations were well represented at all three nest sites but only represented 0.7% of the total recorded point localities for all three nest sites and all were from the Diptank nest site. Orchards and settlements were not represented with any location point data within the viewshed areas.

The individual [Diptank (Χ² = 123.9, df = 5, p < 0.0001), Florida (Χ² = 338.1, df = 6, p < 0.001), Tafeni (Χ² = 168.3, df = 4, p < 0.001)] and combined (Χ² = 601.6266, df = 9, p < 0.001) Chi-square and p-values for habitat preferences were statistically significant.

Of the 11 different land covers represented in the foraging home ranges, 10 were represented in the combined foraging areas, with only the arable habitat class not represented. The total area of the combined areas amounted to 405 ha. Alien vegetation had the highest number of patches (73) but only contributed 8% to the land cover. Grasslands contributed the most to the land cover (49%) with 196 ha. Besides the alien land cover, settlements (0.3%), waterbodies (0.5%) and forests (0.7%) contributed the least. Nests were all located in grasslands and telemetry points congregated around the nest sites (Figures 3 and 4).

The furthest distance travelled by a radio-tagged Blue Swallow was 1296 m and 1030 m, recorded from adult female birds from the Florida and Diptank nests, respectively. It is reasonable to assume that the majority of the foraging activity would take place within 2 km of their nest sites while bound to the nest sites because of chick rearing activities.

Interestingly, the area covered by the adult male Blue Swallow from the Diptank nest appeared to be greater in comparison with the areas recorded for the two female birds. It is known that the male of a species ranges over a larger area than the female (Chandler, Ketterson & Nolan 1997). This increased range could be associated with the male bird seeking extra-pair copulations, which has been noted to be the case in other Hirundines (Turner 2004). Increased movement could be attributed to many factors, including increased testosterone levels, food availability, body mass and population density (Benson, Chamberlain & Leopold 2006; Chandler et al. 1997).

The height of the tagged bird off the ground affected the strength of the radio signal. During the warmer periods of the day, normally between 11:00 and 14:00, the Blue Swallows socialised and flew at higher distances than during active foraging sessions, resulting in much stronger tag signals. However, obtaining reliable signals from afar was limited by topography. This was evident with the positional data obtained from the tagged Tafeni female bird, which frequently disappeared in a southerly direction towards the tea estate where radio tag signal was concomitantly lost. Occasionally, the signal was recovered, but not for long enough periods to obtain an accurate positional fix. This loss of signal owing to terrain is clearly evident in the spatial distribution maps of the Tafeni female, which spent considerable time over the tea plantation during the warmer periods of the day.

The results from the radio tagging showed that commercial timber plantations were clearly avoided by the Blue Swallows. As the majority of these point locations fell just onto the edge of existing plantations, it is more than likely that the points plotted within plantations are possibly a result of error bias during triangulation; an idea supported by the observation that Blue Swallows avoided plantations even though wind direction could have been blowing insects from inside the plantation into the grassland areas (J. Wakelin., pers. obs., November 2004). However, M. McNamara (pers. comm., 2006) has observed Blue Swallows flying along plantation edges in Mpumalanga Province, South Africa. It is likely that Blue Swallows avoid plantations owing to them being aerially cluttered habitats, which structurally do not allow fast-flying aerial foraging birds, such as the Blue Swallows, an unobstructed and safe forage opportunity (Wakelin & Hill 2007).

Alternatively, Blue Swallows, as habitat specialists, could possibly use ecotones within the interior of natural untransformed habitats rather than the periphery of these natural habitats, which have transformed sections. If this is the case, then transformation and habitat fragmentation of the preferred grassland and wetland habitats could have significant conservation implications for the persistence of the species, where fragmentation increases patch number and perimeter length, but reduces available core habitat with suitable interior ecotones (Bruna et al. 2005; Kareiva & Wennergren 1995; Yahner 1997).

Clumps of wattle, Acacia mearnsii, categorised as aliens, were not ignored by the Blue Swallow as a source for insects. The Blue Swallows at the Diptank nest site spent time foraging around small clumps of wattle, and in particular in the lee of these clumps when a wind was blowing. These small clumps of wattle trees were well augmented by indigenous undergrowth. Nevertheless, in time, the situation could change when these wattle clumps might outgrow the protective function that they currently provide to the indigenous species that have established beneath them (Galatowitsch & Richardson 2005).

Waterbodies played a major role in the foraging of the Blue Swallows, which bathed frequently on the wing, almost fully submerging themselves. Faecal sacks were often deposited into the dam by the birds (J. Wakelin, pers. obs., November 2004).

The radio-tagged Blue Swallows spent over 80% of their time foraging over grasslands and wetland habitats and the remainder over tea and sugar cane plantations. This preferential foraging is possibly a result of the increased insect mass and abundance in these habitats (Wakelin 2006). One needs to take cognisance of the fact that grasslands now exist as, in this case, 25 isolated fragments, which could possibly not have been the case in the past, with natural grassland being a single contiguous section interspersed with indigenous forest and wetlands (Camp 1997). These fragments had the third longest perimeter of all the habitats in the study area. Furthermore, these fragments result in extensive perimeter with less suitable habitats and can result in zones of increased threats. It is noteworthy that human-occupied areas were avoided by the Blue Swallows.

Only two land cover classes, other than grassland, are natural: wetland and indigenous forests. As in the case of grassland, natural wetland habitat would have been lost through the construction of dams (Begg 1991), thus reducing a choice habitat for Blue Swallows (Evans et al. 2003; Spottiswoode 2005). Blue Swallows were observed to preferentially use the grassland and wetland habitats’ ecotones as forage zones, more than any other ecotone combination. According to Frouz and Paoletti (2000), insect diversity and abundance is increased on the ecotone between two habitats and may explain this feeding behaviour.

Soil humic content has been suggested to play a key role in determining insect abundance in an area (Callaham et al. 2003; Holland 2004). This may explain the reduced selection by Blue Swallows of the sugar cane fields, which are burnt every 18 months for harvesting, leaving the soil bare and devoid of humus and plant litter. Blue Swallows were observed to forage more frequently and for longer periods over unburnt and ungrazed grassland sections than over recently burnt sections, indicating that fire had possibly negatively affected the quantity of insects available (Callaham et al. 2003; Hanula & Wade 2003).

Loss of suitable natural forage and breeding habitats for the Blue Swallow is cause for serious concern (Evans et al. 2003; Wakelin 2004). Transformed land comprises 71% of the entire combined home range areas. This figure includes 12% of indigenous forest, which has limited use for Blue Swallows as only the ecotones are used for foraging. In effect, there remains intact approximately 29% of grassland and wetland mosaic for the Blue Swallows to breed in and forage on within their active home range areas. Furthermore, increases in fragmentation and perimeter increase the negative edge effect of the transformation (Camargo & Kapos 1995). This in turn reduces habitat availability for the swallows through reducing suitable forage range.

Considering the small and highly fragmented nature of the natural habitats that remain intact, some transformed habitats, such as the tea plantation, are fulfilling an important surrogate role for foraging. Loss of these areas to other land covers less suitable for the Blue Swallow could potentially reduce the availability of forage to less than the critical threshold, which could lead to local extirpation of the species.