© Kamla-Raj 2014 J Life Science, 6(1,2): 13-23 (2014)
An Assessment of Puku (Kobus vardonii) Populations in Kafue National Park, Zambia Using Ground Transects Other than the Road Network
Ngawo Namukonde
School of Natural Resources, Copperbelt University, P.O. Box 21692, Kitwe, Zambia Telephone: +260978695289, Email: [email protected], [email protected]
KEYWORDS Density Estimates. Distance Sampling. Kafue National Park. Puku
ABSTRACT In the past, ground counts in Kafue National Park (KNP) were predominantly confined to the road network owing to the vastness of the Park. Though essential information was yielded, results could not be generalized for the entire area as the road network did not traverse all habitats. Thus, using Distance Sampling methods, an attempt was made in 2010 to obtain unbiased and precise estimates of puku population and structure. Puku was chosen, because it acted as an indicator species for the SEED Programme and secondly its distribution during the dry season was predominantly confined to areas that were relatively accessible. The study area was defined using dry season distributions obtained from daily patrol data and from previous aerial survey reports. With an effort of 152.74km, 90 observations were yielded in groups ranging from 1–76 individuals with a modal size of one. This remained consistent with other studies that have shown that puku usually occur in small groups. Age ratios indicated that there were significantly more adults (81.7%) than sub-adults and juveniles together (18.3%), and may well be a concern to management. Sex ratios were more skewed towards the females, a norm for wild populations. Density estimates (0.433, CV = ± 24.7%) fell within the upper limit of the 2008 aerial count estimate (0.329/km2). Together, these results suggest that Distance Sampling using transects rather than the road network is a more useful and effective technique for estimating densities of puku and should in future be extended to incorporate other large mammals.
INTRODUCTION active adaptive management approach to moni- toring and research of ecological and other pro- From 2006-2011, activities of wildlife research cesses in the park and its surrounding GMAs in the Kafue National Park (KNP) had been (ZAWA 2004). To this end, one of the tasks of steered by the Support for Economic Expansion the SEED Programme in the Kafue National Park and Diversification (SEED) Programme. The (KNP) was to undertake monitoring of animal SEED Programme was a project developed and populations, which helped in determining their implemented by the Government of the Repub- population sizes and demographic status. Aeri- lic of Zambia, through the Ministry of Tourism, al and ground count surveys were the planned Environment and Natural Resources (MTENR) activities in achieving this task. and Zambia Wildlife Authority (ZAWA), with Traditionally ground count surveys have funding assistance from the World Bank, Global been conducted along the road network in KNP, Environment Facility and the Government of particularly in the photographic tourism areas Norway. The goal of the SEED Programme was namely, Busanga; Lufupa; Kafwala; Hippo; to reverse the loss of biodiversity in the park Chunga; Ngoma; and Nanzilla (Bond and Scott and its surrounding Game Management Areas 2003; Miyauchi 1992, 1993, 1994; Mkanda et al. (GMAs), and develop sustainable tourism. The 2005; Mkanda and Simpamba 2005; Simpamba programme was implemented through six com- 2007; Namukonde and Simpamba 2009). These ponents namely: Park administration and man- surveys have provided essential information on agement, Infrastructure development, Resource the procreation potential, encounter rates and protection, Wildlife research and monitoring of densities of several large mammal species. critical habitats, Community based natural re- Though useful, this information could not be sources management, and Private sector-public generalized for the entire Park as the road net- partnership. One objective of Component Four work in KNP does not traverse all habitats. Fur- of the SEED Programme (Wildlife research and ther, some of the species sampled had varying monitoring of critical habitats) was to adopt an habitat requirements and their distribution was 14 NGAWO NAMUKONDE afwala and Chunga Fig. 1. Areas sampled during road counts in KNP 1992-2008. From bottom left to right Nanzhila, Ngoma, Busanga, Lufupa, Hippo, K 1992-2008. From counts in KNP sampled during road Areas Fig. 1. AN ASSESSMENT OF PUKU (KOBUS VARDONII) POPULATIONS IN KAFUE NATIONAL PARK 15 by no means confined to areas along the road tion in the dry season. These areas were defined network that is predominately built parallel to using dry season distribution patterns of puku major rivers and streams. described in the 2008 aerial survey for KNP by Increasing interest in evaluating the effects Frederick (2009) and daily routine patrol data. of management practices on animal populations Figure 2 depicts the distribution of puku in the in KNP led to a need for suitable methods of dry season in KNP. analyzing data generated from ground counts. Thus, the use of the current road network (Fig. METHODOLOGY 1.) as transects for ground counts was deter- mined to afford transects that cut across habi- Puku are predominantly grazers and are al- tats and are positioned according to habitat re- ways found near water (Kingdon 1997; Walker quirements of selected study animals. 1996). Based on this, water ways were utilized as Puku was proposed as the model animal in the basis of selecting transects as opposed to this study, firstly because it was an indicator the road network. Only water ways known to species for the SEED Programme whose popula- harbour water in the dry season were consid- tion dynamics provided an indirect measure of ered whose total length was 2,305 km. From trails the levels of disturbance of their ambient envi- conducted around Chunga area the furthest ronment; and secondly its distribution pattern sighting of Puku from the nearest water way was particularly in the dry season is predominantly less than one kilometre. Therefore, transect confined to open grasslands bordering rivers lengths of 1.5 km and spacing of 17km were pro- and streams (Frederick 2009) that are readily ac- posed. This was done to avoid underestimating cessible. the range if 1km was used and wasting a large The objective of this study was to obtain amount of effort beyond 1.5km. unbiased and precise estimates of puku densi- Equation 7.3 of Buckland et al. (1993) was ties that could be used to monitor trends in their used to estimate values of expected total line population over time. In addition, this study length to be surveyed for a given level of preci- was aimed at determining puku demographic sion, by inputting the calculated encounter rate aspects, that is, cluster sizes, sex ratio and age from trials within Chunga area, a range of values structure that are essential for conservation for the desired coefficient of variation of esti- management (Corti et al. 2002; Bothma and Toit mated density for the Park (that is, 0.1 - 0. 25) 2010), and to test the validity of our assumption and the expected coefficient of variation of clus- that puku occurred at distances less than 1.5km ter size (that is, 0.12 - 0.44). These were calculat- away from water. ed from previous road transects as well as the trial transects. From these values a coefficient Study Area of variation (CV) for density was chosen that is, 0.1 for the desired level of precision and used KNP is located in the centre of south-west- the expected CV of cluster size of 0.3 (being a ern Zambia between 14°03" and 16°43" south fairly conservative estimate) to give an estimat- and 25°13" and 26°46" east in the Central Afri- ed line length to be surveyed of 187.36 km from can Plateau and Miombo Ecoregion (ZAWA which we expected to observe 462 clusters of 2010). It covers an area of approximately 22,480 puku. Given the estimated line length to be sur- km2 and has huge areas of natural habitats with veyed of 187.36 km and the proposed transect extensive woodlands (Brachystegia, Colophos- length of 1.5 km, a total of 127 transects were permum, and Baikiaea), large alluvial flood- required. plains, and a major river (Kafue). One hundred Random numbers tables provided by Sokal and fifty-eight (158) mammals, 481 birds (over and Rholf (1969) were used to randomly select half Zambia’s species, and 80% of all genera), 69 the point from which all transects from the over- reptiles, 36 amphibians, and 58 fish have been all range would be aligned. Thus, the research- recorded in KNP and are distributed throughout ers’ starting point in aligning transects was UTM in varying densities and associations (NPWS/ 378365; 8373730. As it was probable that the JICA 1999). Annual precipitation falls within the density of puku decreased with increased dis- range of 800-1,100mm.The study area was fur- tance from water, transects were placed parallel ther redefined, to include areas of puku distribu- to the direction of the gradient. From these 16 NGAWO NAMUKONDE Fig. 2a. Distribution of puku during the dry season in KNP, 2008 season in KNP, Fig. 2a. Distribution of puku during the dry aerial survey Fig. 2b. Distribution of puku during the dry season in KNP, Patrol season in KNP, Fig. 2b. Distribution of puku during the dry data 2006, 2009 AN ASSESSMENT OF PUKU (KOBUS VARDONII) POPULATIONS IN KAFUE NATIONAL PARK 17 points, the survey team projected way points in culated by fitting detection functions that is, segments of 300m up to 1,500m in the direction half normal, uniform and hazard rate to the data. or bearing to which the transect was placed per- Our final detection model was chosen based on pendicular to the water way. Each transect was the best Chi-squared goodness of fit to the data traversed on foot by a team made up of three and the lowest Akaike Information Criterion members equipped with binoculars, GPS sets, (Buckland et al. 2001; Thomas et al. 2002). Abun- laser rangefinders, angle boards and firearms. In dance was extrapolated by multiplying the den- total there were five teams that would on aver- sity yielded by the total area of the Park age traverse two to three transects a day. All (22,400km2). terrain vehicles were used to get to the start points for each transect. Team member with GPS RESULTS was tasked to keep all members on the transect line and record all observation on a data form A total sampling effort of 152.74km was designed to capture species detected, their num- achieved, against the 187.36km earlier planned. bers and age, sighting distance and angle from The major limitation was access to some of the line, GPS location and habitat type. This team start points that proved to be unreachable us- member was last in line. Ahead was another mem- ing both vehicle and boat. Thus, only 97 ber with an angle board and range finder, whose transects out of the 127 earlier planned were task was primarily to measure sighting distance achieved, which yielded 90 observations of puku and angle from the line. The third team member which is sufficient to undertake Distance analy- carried a firearm and functioned to detect ani- sis (Buckland et al. 2001; Thomas et al. 2002). mals as well as secure the team from danger. Most puku occurred in groups ranging from 1 – 76 individuals with a modal size of 1. This re- Data Collection mained consistent with other studies that have shown that puku are usually found in small The group structure of puku was recorded groups (Corti et al. 2002; Dipotso and Skarpe based on age and sex. Descriptions provided by 2006; Rodgers 1984). Females were more fre- Child and von Richter (1969) and Bothma and quently sighted in groups of more than 6 indi- Toit (2010) were used to age puku as adult, sub- viduals as opposed to the males that were most adult and juveniles or calves. With regards sex, commonly found alone (Table 1). the presence of horns and their stage of devel- opment was used. In both sexes, juveniles had Table 1: Size and sex composition of puku groups no horns and were not sexed. The sub adult counted during the dry season ground counts in males had horns that were not fully developed Kafue National Park and had a straight horn appearance with less Group Frequ- Males Females Juve- lyrate and stem. In adult males, horns were more size ency (%) (%) (%) nile (%) lyrate and curved backwards with their tips 1 33.33 56.92 11.76 16.67 further apart. 2 11.11 15.38 13.73 16.67 The habitat description of each of the sight- 3 4.44 4.62 1.96 25.00 ings was recorded based on the vegetation de- 4 4.44 7.69 9.80 8.33 scriptions of the KNP Vegetation Monitoring 5 1.11 1.54 - 16.67 6-10 18.89 4.62 33.33 8.33 System (Mwima 2006) and the flora descriptions 11-15 10.00 3.08 19.61 - contained in the KNP General Management Plan 16-20 6.67 3.08 3.92 8.33 (NPWS/JICA 1999). The major vegetation types z 10.00 3.08 5.88 - described as occurring within KNP include: Mi- ombo, Acacia, Termitaria, Combretaceae, Ripar- ian, Thicket, Kalahari, Shrubland, Biakaea, Puku were observed in nine out of the elev- Wooded grassland, Grassland and Mopane. en major vegetation types described in KNP Data collected was collated using Microsoft (Mwima 2006; NPWS/JICA 1999). Observations Excel which was also used to undertake basic were most frequently made in wooded grass- analysis. The collated data was then exported to lands and grasslands (26.7%) followed by termi- DISTANCE 5.0 version 2 (Thomas et al. 2006) taria (25.5%), combretaceae (22%), kalahari (20%) that yielded density estimates. These were cal- and Acacia (15.2%) vegetation types. Very few 18 NGAWO NAMUKONDE
1997; Walker 1996). The age ratio proved to be skewed towards the adult populations (Fig. 3). Sub- adult and juvenile population together made up at least 18.3%. Adult female to calf ratio attained was 1: 0.136. With regards to the sex ratio, the study find- ings are that, there is approximately 1 ram for every 2 ewes that is, 1: 1.61 adult male: adult female. Figure 4 illustrates the sex ratio of adult puku populations. The attempt to define the distance at which puku occur from water environments yielded a range of distance from 4 – 1,500m. The modular Fig. 3. Age ratio of puku – 2010 ground counts distance was 100m, while the mean distance was 461.5m. These results were expected based on were sighted in the more closed woodlands that the known habitat requirements of puku that is, 1.72% in riparian, 0.66% in thicket and 0.53% keeps them tied to areas within short walking in miombo. Puku are most frequently described distances to water (Kingdon 1997). It is this as making use of grasslands and open wood- dependency that might have qualified puku pop- lands lying adjacent to water ways (Kingdon ulations as indicators for the management inter- vention of the SEED programme. Half Normal Cosine was the model chosen to define encoun- ter rates and density, based on the low AIC val- ue yielded and that this model ran without warn- ings. The model with the smallest AIC provides, and in some sense represents the best fit of data (Thomas et al. 2009). Secondly this model pro- vided better detection probability curves for pooled data that is, χ2=0.8992, DF=4 P=0.92468. The detection probability curve yielded is illus- trated in Figure 5. A density estimate of 0.433, CV= ±24.7% was Fig. 4. Sex ratio of adult puku – 2010 ground yielded. We considered a 25% CV as relatively counts small and thus estimates yielded are considered
Fig. 5. Detection probability curve for puku – 2010 ground counts AN ASSESSMENT OF PUKU (KOBUS VARDONII) POPULATIONS IN KAFUE NATIONAL PARK 19 as reliable (Buckland et al. 1998). The mean clus- Table 2: Effort, effective strip width, cluster size, ter size (8.60) attained was not significantly dif- density, encounter rates and abundance ferent from those attained in the Chunga (7.35) Parameter Value and Busanga (9.89) areas in the 2008 road count survey (Namukonde and Simpamba 2009). How- Effort 152.74 ever, our estimated average cluster size was way Effective Strip Width (m) 160.00 Encounter rate 0.59 below the amalgamation numbers of puku (>50) Encounter rate CV 0.15 described by Kingdon (1997) in the dry season. Mean Cluster Size 8.60 Despite this, the estimate fell within the range of Mean Cluster Size (CV) 0.14 the wet season amalgamation numbers of 3-15 Mean Cluster Size Lower Confidence 6.46 Limit (DLCL) individuals. The encounter rate yielded (0.59) Mean Cluster Size Upper Confidence 11.45 was significantly lower than those of the 2007 Limit (DUCL) (4.90) and 2008 (1.07) road counts (Simpamba Density of individuals (per km2) 0.43 2007; Namukonde and Simpamba 2009). Table 2 Density of individuals CV (per km2) 0.25 provides a summary of the results obtained from Density of individuals LCL (per km2) 0.32 Density of individuals UCL (per km2) 0.54 Conventional Distance Sampling Engine.
DISCUSSION Kilombero Valley (Corti et al. 2002; Dipotso and Skarpe 2006; Waltert et al. 2009). In these areas it The study’s results suggest that the sam- was found that puku did not generally form large pling methods used in this study are adequate groups like other species of Kobus that is, red for estimating densities of puku and can be used lechwe. Overall sex ratio yielded (1 adult male to monitor populations over a period of time. :1.61 adult females) is described as normal for Use of transects instead of the road network wild populations (Bothma and Toit 2010) and afforded random transects, a major assumption compares well with those described in Kasanka in Distance Sampling (Thomas et al. 2009) which National Park of 1: 1.56 and the Chobe River yielded a more unified distribution of observa- Front of 1: 1.5–1.6 (Goldspink et al. 1998; Child tions. This was previously impossible when the and von Richter 1969). road network was used, as transects followed The age ratios yielded allude to an ageing roads and tracks that are predominately built population of puku in KNP, as over 80% of the parallel to rivers and streams. Thus, poor detec- populations are adults. Together, juveniles and tion probability curves were afforded for spe- sub adults made up at least 18.3%. Whether, cies like puku (Namukonde and Simpamba 2009) this should be an immediate management con- whose habitats were confined to these areas (Fig. cern, requires an all year round assessment, as 6). It was only in the Chunga area where rela- puku are known to calve all year round, with tively good shoulders were yielded for puku. peaks in the wet season (Kingdon 1997; Dipot- Similar shoulder patterns were also yielded so and Skarpe 2006). This study was confined only to the dry season that is, October 2010. by Bond and Scott (2003) for puku and water- Adult female to calf ratio attained (1: 0.136) fell buck in Ngoma (an area in the southern part of below other puku populations described by Di- KNP) where roads were used as transects dur- potso and Skarpe (2006) of 1:0.37 and Child and ing their large mammal census. This again did von Richter (1969) of 1: 0.20. The adult female to not quite meet the primary assumption of Dis- calf ratio for puku is known to vary consider- tance Sampling, that all objects of interest are ably between months with peaks in the early detected on the line with certainty (Thomas et wet season (Kingdon 1997). This may explain al. 2009). the disparity observed as sampling was con- Understanding the group sizes in herbivores ducted in the late dry season of 2010. is cardinal in conservation management as it pro- With eleven observations beyond 1,000m, the vides ecologists with basic understanding of transect length may have been underestimated. savannah communities (Corti et al. 2002). Thus, However, for some of these observations (27%), group sizes were examined in this context and it several pools of water that were not expected has been revealed that the group sizes of puku were encountered along these transects. Fur- in KNP compare very well, with those of the ther, observations less than 1,500m have been Chobe River Front, Lake Rukwa water front, and described as frequent in similar studies (Waltert 20 NGAWO NAMUKONDE
Fig. 6(i) Detection probability curve for UC AIC model (Ngoma – 2008 road counts)
Fig. 6(ii) Detection probability curve for HRC AIC model (Busanga– 2008 road counts)
Fig. 6(iii) Detection probability curve for UC AIC model (Hippo– 2008 road counts) AN ASSESSMENT OF PUKU (KOBUS VARDONII) POPULATIONS IN KAFUE NATIONAL PARK 21
Fig. 6(iv) Detection probability curve for UC AIC model (Kafwala–2008 road counts)
Fig. 6(v) Detection probability curve for HRC AIC model (Lufupa– 2008 road counts)
Fig. 6(vi) Detection probability curve for HRC AIC model (Chunga-2008 road counts) Fig. 6. Detection probability curves, yielded in the six areas sampled during the 2008 road counts 22 NGAWO NAMUKONDE
et al. 2009). Needless to say, the transect lengths Child G, von Ricther W 1969. Observation on the ecol- should in future be lengthened to at least 4 - ogy and behaviour of lechwe, puku and waterbuck along the Chobe River, Botswana. Zschr Sgtk, 34: 10km which Tainton (1999) described as distance 275–295. grazers will move from surface water to graze. Corti G, Fanning E, Gordon R, Hinde J, Jenkins RKB 2002. Observations of puku antelope (Kobus vardo- nii Livingstone 1857) in Kilombero Valley, Tanza- CONCLUSION nia. Afr J Ecol, 40: 197-200. Dipotso FM, Skarpe C 2006. Population status and The results of this study demonstrate that distribution of puku in a changing riverfront habitat ground counts using transect instead of the road in northern Botswana. S Afr J Wildl Res, 36: 89–97. Frederick H 2009. Aerial Survey of Kafue Ecosystem network in KNP Park can yield precise and reli- 2008. Lusaka: Zambia Wildlife Authority. able estimates. This is corroborated by the sim- Goldspink CR, Holland RK, Sweet G, Stjernstedt R 1998. ilarity in the results obtained, between those of A note on the distribution and abundance of puku, similar studies conducted within the region. Kobus vardonii Livingstone, in Kasanka National These results provide a basis for further im- Park, Zambia. Afr J Eco, 36: 23-33. Kingdon J 1997. The Kingdon Guide to African Mam- provement of ground based population studies mals. London: A and C Black. in KNP that will in future provide more cost ef- Miyauchi F 1992. Large Mammals Monitoring at Prin- fective tools to evaluate effects of management cipal Areas in the Kafue National Park–Area III: practises on animal populations. Treetops (with Busanga Plains). Lusaka: National Parks and Wildlife Services. Miyauchi F 1993. Large Mammals Monitoring at Prin- RECOMMENDATIONS cipal Areas in the Kafue National Park–Area II: Nanzhila. Lusaka: National Parks and Wildlife Ser- The sampling methods used in this study vices. Miyauchi F 1994. Large Mammals Monitoring at Prin- should be extended to incorporate other large cipal Areas in the Kafue National Park–Area IV: mammal species within KNP. Studies aimed at Lufupa. Lusaka: National Parks and Wildlife Servic- understanding the factors responsible for the es. age structure observed in this paper is also Mkanda FX, Phiri HA, Zyambo L 2005. Preliminary recommended. Results of a Road Count of Mammals in the Kafue National Park (North). Mumbwa: Zambia Wildlife Authority. ACKNOWLEDGEMENTS Mkanda FX, Simpamba T 2005. Preliminary Results of a Road Count of Mammals in the Kafue National Gratitude is extended to the cooperating part- Park. Mumbwa: Zambia Wildlife Authority Mwima H 2006. Vegetation Monitoring Manual for the ners under the SEED Programme that is, World Kafue National Park. Mumbwa: Zambia Wildlife Bank and Government of Norway that provided Authority. support for research activities in the KNP. I am Namukonde N, Simpamba T 2009. A Technical Report also thankful to the KNP staff that contributed of Road Counts of Large Mammals in the Kafue significantly towards data collection and entry. National Park. Mumbwa: Zambia Wildlife Authority NPWS/JICA 1999. General Management Plan for Kaf- Special thanks go to Janice May and Rolf Lind- ue National Parks and its Surrounding Game Man- holm (sitatunga researchers), whose patience agement Areas. Lusaka: National Parks and Wild- and invaluable knowledge on the topic of study life Services. made all the difference. Rodgers WA 1984. Status of puku (Kobus vardoni Liv- ingstone) in Tanzania. Afr J Ecol, 22:117–125. Simpamba T 2007. 2007 Road Count Report- Kafue REFERENCES National Park. Mumbwa: Zambia Wildlife Authority. Sokal RR, Rholf FJ 1969. Statistical Tables. San Fran- Bond G, Scott DM 2003. Mammal Density Estimates– cisco: W. H. Freeman and Company. Distance Sampling. Itezhi Tezhi: Zambia Wildlife Tainton NM 1999. Veld Management in South Africa. Authority and GREENFORCE. Pietermaritzburg: University of Natal Press. Bothma J du P, Toit du JG 2010. Game Ranch Man- Thomas L, Buckland ST, Burnham KP, Anderson DR, agement. Pretoria: Van Schaik. Laake JL, Borchers DL, Strindberg S 2002. Distance Buckland ST, Anderson DR, Burnham KP, Laake JL sampling. In: Abdel H El-Shaarawi, WW Piegorsch (Eds.): Encyclopedia of Environmetrics. Chichester: 1993. Distance Sampling: Estimating Abundance John Wiley and Sons, pp. 544–552. of Biological Populations. London: Chapman and Thomas L, Buckland ST, Rexstad EA, Laake JL, Strind- Hall. berg S, Hedley SL, Bishop JRB, Marques TA, Burn- Buckland ST, Anderson DR, Burnham KP, Laake JL, ham KP 2009. Distance software: Design and analy- Borchers DL, Thomas L 2001. Introduction to Dis- sis of distance sampling surveys for estimating pop- tance Sampling. Oxford: Oxford University Press. ulation size. J Appl Eco, 10: 1365-2664. AN ASSESSMENT OF PUKU (KOBUS VARDONII) POPULATIONS IN KAFUE NATIONAL PARK 23
Distance 5.0. Release “2” 2006. From