«Assessing the status of Handsome Francolin Francolinus nobilis in Bwindi Impenetrable National Park, western Uganda Richard Ssemmanda and Richard A. ...»
Scopus 25: 4150, December 2005
Assessing the status of Handsome
Francolin Francolinus nobilis in Bwindi
Impenetrable National Park, western
Richard Ssemmanda and Richard A. Fuller
The Handsome Francolin is a ground-dwelling partridge occurring in
montane forest and the high-altitude bamboo zone within a restricted area
along the Albertine Rift Mountains from the Bleus Mountains in eastern
Democratic Republic of the Congo (DRC) south to Burundi (Urban et al.
1986, Madge & McGowan 2002). It appears to occur over a global range of about 120 000 km2 (Fuller et al. 2000, BirdLife International 2004), although there are very few recent records from most parts of this area.
The small area of the Albertine Rift Mountains supports 41 endemic bird species, 11 of which are globally threatened (BirdLife International 2004, Plumptre et al. 2003). Forest covers much of the 56 000 km2 in the Albertine Rift Endemic Bird Area, and much of it is rugged terrain very difficult to access (Shaw & Shewry 2001). Despite this, there are significant human pressures in the area, stemming mostly from large increases in human population densities as a result of refugee movements (Omari et al. 1999, Gatarabirwa et al. 2000). Being large, slow-moving, palatable terrestrial birds, the Galliformes are under particularly direct pressure from humans through hunting and disturbance, and they may provide useful rapid indicators of the amount of human pressure in a particular area.
The area surrounding Bwindi is one of Ugandas most densely populated rural areas with human densities of 160320 people/km2. Approximately 10 000 families cultivate the land immediately surrounding the park (Butynski 1984). About 84 % of the forest compartments display signs of human activity, including pit sawing (29 %), hunting (24 %), mining (6 %), livestock (10 %) and footpaths (67 %) (Butynski & Kalina 1993). This suggests that species providing consumable bush meat are at particular risk.
Pairs and small groups of Handsome Francolins have been encountered in southwestern Uganda (Prigogine 1971, Dehn & Christiansen 2001), and it is frequently seen by birdwatchers, though many visitors see the species within a very small area near Ruhiza, in Bwindi Impenetrable National Park (Rossouw & Sacchi 1998). It has been described as locally common within its small global range (Fuller et al. 2000), although there has never been any scientific survey to measure its abundance. No research has ever been Richard Ssemmanda and Richard Fuller 42 conducted on this species; its nest and eggs are undescribed, its diet and breeding season are unknown and, in view of possible threats, research on its ecology is now urgently needed (Dehn & Christiansen 2001).
As an illustration of how difficult this species is to detect in the field, recent extensive bird surveys using circular plot methods in Bwindi Impenetrable National Park, Uganda, detected Handsome Francolins only once (Shaw & Shewry 2001), and intensive fieldwork in Rwenzori Mountains National Park in 1996 found the species at only one location (Dehn & Christiansen 2001). Handsome Francolins are only seen with any frequency when foraging on wide roads that run though the bamboo zone near Ruhiza (Rossouw & Sacchi 1998) and rarely call in natural situations.
Playback surveys have been successfully used to census populations of elusive birds including several species of Galliformes (Glahn 1974, Marion et al. 1981, Gibbs & Melvin 1993, Fuller et al. 2004). Playback surveys often detect more birds than conventional methods (Sliwa & Sherry 1992) and have been used to study globally threatened species (Njoroge & Bennun 2000, Carroll & Hoogestein 1995). So we set out to design and test a survey method capable of producing a density estimate for the Handsome Francolin. One possible way of sampling the birds is to play a recording of the advertising call and use the distribution of distances between the observer and the responding bird to generate a detection function (Bibby et al. 2000, Buckland et al. 2001). We adapted a playback technique used to census Nahans Francolins Francolinus nahani in Uganda (Sande 2001, Fuller et al. 2004).
In addition to this, we present new data on altitudinal range, broad habitat preferences and the distribution of the species within Bwindi Impenetrable Forest National Park.
Study area and methods Between January and March 2004 we studied Handsome Francolins in Bwindi Impenetrable Forest National Park, western Uganda (00º03S, 29º40E). With an area of 32 092 ha, the park is completely protected on paper, although extractive use is occasionally sanctioned and illegal extraction of forest resources does occur. Steep hills and narrow valleys characterize Bwindi with a general incline from the northern and western areas below 1750 m, to the southwestern corner above 2250 m. Bamboo thickets are restricted to less than 100 ha. The area is broadly classified as medium-altitude moist evergreen and high-altitude forest (LangdaleBrown et al. 1964). Bwindi is an important locality for the conservation of Afromontane fauna in particular those endemic to the Albertine Rift Mountains (Plumptre et al. 2003). See Howard (1991) for further details of vegetation types and climatic conditions within the park.
The observer (RS) walked along routes of varying length through the reserve (Figure 1), establishing survey points at least 200 m apart along each Handsome Francolin in Bwindi, Uganda 43 route. Direct distances between points were measured using a handheld GPS receiver rather than distance along the walked route. At each point, the advertising call of the Handsome Francolin (recording taken from Chappuis
2001) was played using a Nakai tape recorder and Sony battery-powered 10 W speaker.
Figure 1. Map showing the location of Bwindi-Impenetrable National Park (shaded grey) in southwest Uganda, together with locations of the survey routes (shown in black).
The Handsome Francolin advertising call was played for 20 s, and any response noted in the ensuing 120 s. This process was repeated twice, to give a total of three successive playbacks. The distance to any calling group, and the playback episode after which it responded were noted for each responding group. Distances were estimated to the nearest 15 m. The observer had extensive previous experience of estimating distances to calling birds, and prior to commencing fieldwork, checked the accuracy of distance estimation by visually locating calling birds along tracks around Ruhiza and confirming their actual distance.
All playback surveys were conducted between 07:00 and 13:00, as the birds appeared to call more frequently in the morning. In the case of rain (only one day was affected), surveying was suspended until the rainfall had ceased. Routes were walked entirely along existing trails in the forest. The reasons for this were fourfold. Firstly, we wanted to minimise the impact of Richard Ssemmanda and Richard Fuller 44 our work, and opening up new trails through the forest would have caused environmental damage as well as opened up the areas for hunters and loggers to enter the forest. Secondly, cutting our own path through the forest would have made a lot of noise, and probably seriously affected the results of the survey if birds were repelled from the point count area.
Thirdly, cutting trails is time-consuming and would have severely reduced our sample sizes and the comprehensiveness of our survey work. Fourthly, the terrain was extremely steep and precarious in places, so for safety reasons a strictly randomised design was not possible. Most of the trails were very narrow and unlikely to bias the results through edge effects.
A handheld GPS was used to measure location and altitude at each survey point with an average error of about 10 m. Broad habitat type at each survey point was classified as forest, bamboo or a mixture of the two. To investigate habitat selection by the francolins in more detail, a sample of points where (a) francolins were detected and (b) not detected, was chosen at random along each route. Points were chosen to provide an approximately equal sample of locations at which francolins were detected and locations at which francolins were not detected. To avoid disturbing birds, habitat surveys were conducted after playback sampling was completed each day, typically by retracing our steps along the route and stopping at the randomly selected points to take habitat measurements.
At points where francolins were detected, the estimated distance and measured bearing were used to locate the approximate position of calling groups, from where detailed habitat data were collected. At points where francolins were not detected, habitat data were collected 30 m from the survey point in a random direction chosen using random number tables.
Where it was physically impossible to reach the position of a calling francolin group (for example where the terrain was too steep to be crossed safely) or the randomly selected position, detailed habitat data were not collected at that point.
For detailed habitat data, a 5 x 5 m quadrat was estimated around the sample location. The observer stood at the centre of this quadrat and estimated the following habitat variables: (i) height of the canopy, defined as all vegetation above 11 m in height; (ii) percentage canopy cover (iii) height of the understorey, defined as vegetation between 2.5 and 11 m in height; (iv) percentage cover by the understorey; (v) height of the ground vegetation, defined as all vegetation less than 2.5 m in height;
(vi) percentage cover of ground vegetation (vii) circumferences of all live trees within the quadrat with a diameter at breast height (dbh) 20 cm;
(viii) number of trees with dbh 20 cm; (ix) depth of leaf litter, expressed as the average leaf litter depth at four random points in the quadrat;
(x) distance to the nearest buttressed tree (possibly used as nesting or roosting sites, as in other forest-dwelling francolins), defined as a tree with at least one buttress that was distinct from the trunk at 2 m from the Handsome Francolin in Bwindi, Uganda 45 ground; (xi) circumference at breast height of the nearest buttressed tree;
(xii) number of buttresses on the nearest buttressed tree. If a buttressed tree could not be located within 30 m of the centre point, the search was abandoned.
Results Playback surveys Between January and March 2004, playbacks were conducted at 244 points.
Responses by 25 groups at 21 points were obtained, giving a ratio between the number of points surveyed to responding groups of 10:1. Handsome Francolins rarely called spontaneously; francolin calls not stimulated by tape playback were heard only nine times during the fieldwork period.
Most of these spontaneous calls were heard around dawn and dusk, (05:00 06:30 and 19:0019:45) and were in or near the bamboo zone at Ruhiza. Our dataset was not large enough to generate a robust population density estimate, but we have succeeded in demonstrating that playback can be used to survey for Handsome Francolins.
Elevation The altitude of survey points ranged from 1468 m to 2541 m. Handsome Francolins were detected between the altitudes of 1541 m and 2541 m. The altitude of survey points at which birds responded to the tape was higher than the altitude of survey points where no response was detected (median altitude with no response = 2200 m, median altitude with response = 2399 m; U223,21 = 553.5, p = 0 011), suggesting that birds were selecting higher altitude regions of the study area.
Broad habitat type The broad habitat types surveyed were altitudinally distinct. Forest occurred from the lowest parts of the reserve at 1500 m up to 2200 m, and the bamboo zone from about 2400 m upwards. The vegetation between these two zones was mixed.
Francolins were detected far more frequently in bamboo habitat than in forest habitat (Gadj = 13.018, p 0.001). Birds responded at over 30 % of survey locations in bamboo and about 6 % of the locations in forest, while there were no responses in any of the mixed habitat survey points (Table 1).
Habitat structure A total of 50 habitat plots was assessed, comprising 31 plots at randomly chosen points where francolins were not heard, and 19 at points where francolins were heard.
Canopy cover or height did not differ between random and francolin
survey points (canopy cover: t = 0.01, d.f. = 48, p = 0.992; canopy height:
t = 0.82, d.f. = 48, p = 0.416). However, variables relating to the structure of Richard Ssemmanda and Richard Fuller 46 Table 1. The number of survey points in each of the six broad habitat types where francolin groups responded to playback. Because no birds responded in mixed forest, data were analysed using a G-test with Williams correction on a 2 x 2 contingency table to compare the frequency of detections in bamboo and forest habitat types.
Bamboo Forest Mixed Total Francolins not detected 19 183 21 223 Francolins detected 9 (32%) 12 (6%) 0 (0%) 21 (9%) Total 28 195 21 244 the understorey and ground vegetation were strong predictors of francolin presence. Understorey was denser at locations where francolins were found than at random points (t = 2.1, d.f. = 48, p = 0.044), and percentage cover of ground flora was much lower (t = 2.68, d.f. = 48, p = 0.010), suggesting that birds preferred areas with more bare ground, but denser understorey cover.
Tree density was much higher at locations where francolins were detected than at random points (francolin points = 35 trees per quadrat; random points = 19 trees per quadrat; t = 3.66, d.f. = 48, p 0.001). Furthermore, the girth large of trees (those with a dbh greater than 20 cm) was significantly higher at francolin points than at random points (francolin points = 1.55 m;
random points = 2.08 m; t = 2.67, d.f. = 48, p = 0.010).
Distance to the nearest buttressed tree and the number of buttresses on the nearest buttressed tree did not differ between random and francolin survey points (distance: U21,15 = 121.0, p = 0.252; number of buttresses: U23,17 = 194.5, p = 0.978). The circumference of the nearest buttressed tree was significantly larger at locations where francolins were found than at random points (U21,15 = 76.0, p = 0.008), although this probably simply reflects the general association with forest comprising large trees described above.