FIXED-WING AERIAL SURVEY OF WILDLIFE IN THE ABU WILDLIFE MANAGEMENT AREA (NG 26) OKAVANGO DELTA, BOTSWANA OCTOBER 2010 Report Prepared by Michael Chase PO Box 682 Kasane Botswana Tel: ++267 6250505 Email: er@info.bw Moremi Game Reserve January 2011
ACKNOWLEDGMENTS This aerial survey of wildlife over the Abu concession (NG 26) was commissioned and funded by Elephant Back Safaris (Pty) Ltd. The survey over part of Habu (NG 8) was funded by the Department of Wildlife and National Parks (DWNP) and the San Diego Zoo. I thank Mr. Mike Holding of Afriscreen Films for flying this survey. The DWNP are acknowledged for their ongoing support of this work. I appreciate the assistance of our aerial observers, Ms. Kelly Landen and Mr. Adrian Dandridge. I am grateful to Kelly Landen who counted and verified wildlife numbers from the photos taken by observers during the survey. I am grateful to Ms. Tania Jenkins from Afriscreen Films for providing administrative assistance during this survey. The management and staff of Seba Camp are thanked for their hospitality and logistical support. I thank Ms. Cally Hanley and Mr. Adrian Dandridge for providing me with the opportunity to conduct this important aerial survey over the Abu concession (NG 26). Mr. Mike Holding is acknowledged for reviewing an earlier draft version of this report. Cover Photo: Kelly Landen, 2010. A family group of elephants drink from a floodplain in the Abu concession. ii
EXECUTIVE SUMMARY During October 2010, a fixed-wing aerial wildlife survey was flown in the Abu Elephant Back Safari concession (NG 26), and part of NG 8 which borders NG 26 in the Okavango Delta. This wildlife aerial survey was commissioned by Elephant Back Safaris (Pty Ltd). The survey area sampled totaled 2338 km 2. The principal objective of this survey was to provide relatively precise and accurate estimates of the numbers of wildlife in the survey area. This is the second survey which has been flown over the concession. This report provides the results of this survey, in addition to information on the spatial distribution, abundance and trend of wildlife species in the concession. The survey area was divided into four strata. The Abu concession was divided into three strata, within each stratum the transects were parallel and regularly spaced 1 km apart. We increased the sampling intensity to ~ 40 % compared to an earlier survey (~ 20 % coverage) to improve the precision of our population estimates. Within the Habu stratum transects were spaced 2 km apart. Overall mean search effort was 1.2 minutes per km 2. Aerial surveys often underestimate wildlife numbers, with the degree of underestimation higher for small or cryptic species than for large species. High resolution digital cameras provided photos to compensate for any underestimating or missed animals. The locations of wildlife herds seen during the survey were entered into a GIS to produce maps showing the spatial distribution and herd sizes of principal large herbivores and birds in Abu. There were an estimated 17339 cattle in the Habu tribal agricultural lands bordering Abu, which are separated from the concession by the Southern Buffalo Fence. This largely subsistence farming and tribal cattle grazing area has a large and valuable zebra population (2093). Wildlife in the Habu tribal area is separated from the Okavango Delta by the Southern Buffalo Fence. Programmes to conserve this regions wildlife and unique habitats are urgently needed. iii
The estimated population numbers for the principal large herbivores and birds in Abu concession were: Species Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) Woodland Sp Elephant 3606 1453 81527 16 3042 4169 1.95 Elephant Bulls 531 213 2364 18 435 627 0.29 Elephant Family 3075 1240 76238 18 2528 3621 1.66 Ele Carcass 3 65 25 104 31 44 85 0.04 Buffalo 3745 1523 740163 45 2043 5446 2.02 Giraffe 426 173 3980 29 310 550 0.23 Impala 2705 1094 176939 31 1873 3537 1.46 Kudu 308 125 2051 29 218 397 0.17 Tsessebe 278 111 2352 34 182 374 0.15 Warthog 131 51 427 31 90 171 0.07 Wildebeest 151 57 2407 64 57 248 0.08 Zebra 1303 524 57010 36 831 1775 0.70 Wetland Sp Hippo 287 117 3055 38 177 396 0.15 Reedbuck 58 24 156 41 33 83 0.03 Lechwe 1995 811 47744 22 1563 2427 1.08 Bird Sp Bateleur eagle 94 47 258 66 62 126 0.05 Ostrich 97 35 1058 66 35 161 0.05 Saddle-bill stork 154 61 449 27 112 196 0.08 Wattled crane 148 59 767 36 93 202 0.08 Fish eagle 146 59 524 31 101 191 0.08 1 Estimates for wildlife and cattle seen in Habu are provided in Tables 8 to 24. The report also compares the results of this survey to the first fixed-wing aerial survey conducted in September 2009 (Sept09). Differences between the two fixed-wing aerial surveys (Sept10 & Oct09) of wildlife in the Abu concession are also presented and compared. While the surveys provide a good overview of the sizes of wildlife populations, there is still insufficient information to say much about trends other than that there are no population crashes. Maps, and tables illustrating the distribution, numbers and density of wildlife species in the concession are provided. iv
TABLE OF CONTENTS ACKNOWLEDGMENTS...ii EXECUTIVE SUMMARY...iii TABLE OF CONTENTS... v LIST OF TABLES... vi LIST OF FIGURES...viii Page FIXED-WING AERIAL WILDLIFE SURVEY OF ABU ELEPHANT BACK SAFARI CONCESSION (NG 26) OKAVANGO DELTA, BOTSWANA... 1 Introduction... 1 Study Area... 2 Methods... 4 Fixed-Wing Aerial Survey... 4 Flight Procedures... 4 Observations... 6 Strip Width and Calibration... 7 Data Analysis... 8 Strip Transect Sampling / Fixed-Wing Aerial Survey... 8 Search Effort... 9 Results... 10 Sampling Effort... 10 Wildlife Estimates... 11 Observations... 13 Wildlife Distributions... 16 Habu (NG8) Survey... 17 Trends in Wildlife Numbers in Abu... 19 Discussion... 20 Wildlife Estimates... 20 Observations... 20 Wildlife Distribution... 21 Trends in Wildlife Numbers... 21 Recommendations... 22 Conclusion... 22 References... 24 Appendix 1... 44 v
LIST OF TABLES Table Page Table 1. Sampling statistics for the aerial survey of wildlife in Abu concession, Oct10... 10 Table 2. Wildlife estimates and statistics for major wildlife species, elephant carcasses, baobab trees and birds in Abu 1, NG 26 WMA during aerial survey, Oct10.... 12 Table 3. Numbers seen, herds observed and average herd size of animals, baobabs and carcasses seen in each stratum.... 13 Table 4. Comparison of numbers of individual animals seen and numbers of herds/groups seen by the left and right observers... 14 Table 5. Comparison between observer estimates and photographs of the same animals.... 15 Table 6. Number, size and the extent of damage to baobab trees seen during the aerial survey... 16 Table 7. Estimated numbers, densities and differences between estimated numbers of animals Sept09 and Oct10, aerial surveys... 19 Table 8. Population estimates and statistics for elephants in Abu (NG 26) and Habu... 26 Table 9. Population estimates and statistics for elephant bulls in Abu (NG 26) and Habu... 27 Table 10. Population estimates and statistics for elephant family groups in Abu (NG 26) and Habu... 28 Table 11. Estimates of elephant carcasses in Abu (NG 26) and Habu,.... 29 Table 12. Population estimates and statistics for buffalo in Abu (NG 26) and Habu... 30 Table 13. Population estimates and statistics for giraffe in Abu (NG 26)... 31 Table 14. Population estimates and statistics for hippopotamus in Abu (NG 26) and Habu.... 32 Table 15. Population estimates and statistics for impala Abu (NG 26) and Habu.... 33 vi
Table 16. Population estimates and statistics for kudu Abu (NG 26) and Habu.... 34 Table 17. Population estimates and statistics for lechwe Abu (NG 26),.... 35 Table 18. Population estimates and statistics for ostrich in Abu (NG 26) and Habu.... 36 Table 19. Population estimates and statistics for reedbuck Abu (NG 26) and Habu... 37 Table 20. Population estimates and statistics for tsessebe Abu (NG 26).... 38 Table 21. Population estimates and statistics for wildebeest Abu (NG 26).... 39 Table 22. Population estimates and statistics for zebra Abu (NG 26), and Habu.... 40 Table 23. Population estimates and statistics for birds Abu (NG 26),... 41 Table 24. Population estimates and statistics for cattle in Habu (NG 8),.... 43 vii
LIST OF FIGURES Figure Page Figure 1. Abu WMA (concession number, NG 26) in the Okavango Delta, northern Botswana...2 Figure 2. Abu fixed wing aerial survey strata determined by major land cover types 1...3 Figure 3. The distribution of all wildlife seen during the aerial survey of Abu (NG 26), Oct10....17 Figure 4. Distribution of cattle and wildlife in Abu (NG 26) and Habu (NG 8)....18 Figure 5. Land use map and key features in and surrounding Abu (NG 26) WMA...25 Figure 6. Distribution of elephants in Abu and Habu, Oct10...26 Figure 7. Distribution of elephant bulls in Abu (NG 26) and Habu, Oct10....27 Figure 8. Distribution of elephant family groups in Abu (NG 26) and Habu, Oct10...28 Figure 9. Distribution of elephant carcasses (category 3) in Abu (NG 26) and Habu, Oct10...29 Figure 10. Distribution of buffalo herds in Abu (NG 26) and Habu, Oct10....30 Figure 11. Distribution of giraffe in Abu (NG 26), Oct10....31 Figure 12. Distribution of hippopotamus in Abu (NG 26) and Habu, Oct10...32 Figure 13. Distribution of impala Abu (NG 26) and Habu, Oct10...33 Figure 14. Distribution of kudu Abu (NG 26) and Habu, Oct10...34 Figure 15. Distribution of lechwe Abu (NG 26), Oct10...35 Figure 16. Distribution of ostrich in Abu (NG 26) and Habu, Oct10...36 Figure 17. Distribution of reedbuck Abu (NG 26) and Habu, Oct10....37 Figure 18. Distribution of reedbuck Abu (NG 26), Oct10...38 Figure 19. Distribution of wildebeest Abu (NG 26), Oct10....39 viii
Figure 20. Distribution of zebra in Abu (NG 26), and Habu, Oct10...40 Figure 21. Distribution of birds in Abu (NG 26) Oct10....41 Figure 22. Distribution of cattle in Habu (NG 8), Oct10...43 ix
FIXED-WING AERIAL WILDLIFE SURVEY OF ABU ELEPHANT BACK SAFARI CONCESSION (NG 26) OKAVANGO DELTA, BOTSWANA Introduction This survey was flown in October 2010 (Oct10) and is the second fixed-wing aerial wildlife survey of the Abu Wildlife Management Area (WMA, concession number - NG 26) in the Okavango Delta. The first survey was conducted in September 2009 (Sept09) (Viljoen 2009). For the Oct10 survey, the area sampled was increased to include part of NG 8, a tribal farming area which borders NG 26. Both surveys were flown by means of a sample count. Monitoring large herbivores is central to research and management activities in many conservation areas. Aerial surveys were originally developed to estimate (trends in) population sizes of individual species. However, emphasis is shifting increasingly towards conservation of diversity and communities instead of individual species, as a growing literature shows the importance of herbivore diversity for ecosystem functioning (Joris et al. 2008). Large woodland mammals (elephant, buffalo, wildebeest, zebra and impala), wetland species (hippo) and floodplain ungulates (lechwe, reedbuck, and waterbuck) were counted. Large birds (cranes, storks, ground hornbill, fish eagle and bateleur) were also recorded during the survey. This report presents information from the Oct10 survey but also compares data with the first aerial survey flown in Sept09. These surveys provide valuable information and afford resource managers with the opportunity to assess wildlife distribution, abundance and trends between aerial surveys, and compare changes over time. These data also give the Department of Wildlife and National Parks (DWNP) and the Abu concession important information to better conserve and manage their wildlife populations. The survey also contributes to comanagement of wildlife between protected areas, tribal lands and adjoining ecotourism concessions. This co-management responsibility for monitoring wildlife aims to strengthen DWNP aerial surveys and other wildlife monitoring programmes. Higher order goals and objectives for wildlife conservation in the Okavango Delta are best achieved through such private sector collaboration and responsibility. At a larger scale the surveys contribute important data to current conservation and development initiatives such as the Okavango Delta Management Plan, Bio-Okavango and independent wildlife research projects. The survey area is also part of the Kavango Zambezi Transfrontier Conservation Area and Ramsar Site. The Abu concession, borders Moremi Game Reserve (GR), and is a critical part of the Okavango Delta ecosystem, that provides linkages and corridors for wildlife movement (Chase 2009). 1
Study Area The Abu concession (~ 1853 km 2 ) is a photographic non-consumptive WMA (concession number - NG 26), situated in the upper-central portion of the Okavango Delta in northern Botswana. Prior to 2009, big game sport hunting occurred in the western half of the concession, this activity was stopped in October 2008. NG 26 is now solely used for photographic safaris. The concession has three permanent camps. Abu Elephant Camp is the main base of operation and tourist lodge. Located within 5 km of Abu is its small sister camp, Seba Camp. African Horse Back tented camp is located 12 km southeast of these two camps. To the north, NG 26 is bordered by WMA NG 25. The Jao River (which flows into the Boro R) forms the concessions northeast boundary separating the concession from the Moremi GR. The NG 27A cut line which joins the Jao River then forms the southeast boundary. A similar WMA, NG 29, forms the southern border. In this concession both photographic and hunting are permitted. Tribal farming land, NG 8, is situated in the west, and is separated from NG 26 by the Southern Buffalo Fence. The Potae River meanders into NG 26 for ~10 km in the southwest corner of the concession. The Thaoge River flows outside of NG 26, ~12 km west of the Southern Buffalo Fence through the middle of NG 8 (Figure 1). Figure 1. Abu WMA (concession number, NG 26) in the Okavango Delta, northern Botswana. Okavango R Northern Buffalo Fence Chobe NP Okavango Delta Boro R NG 25 Moremi GR Thaoge R NG 8 Abu - EBS NG 26 NG 27a NG 29 Southern Buffalo Fence 2
The vegetation in the concession can be categorized into three main habitat types: the perennial swamp area, seasonally inundated floodplain grassland, and drier vegetation in the south (Bekker & de Wit 1990). A total of six basic vegetation types have been identified for the survey area, primarily based on satellite imagery (HOORC): Dry floodplain and island interiors, swamp vegetation, shrub grassland on seasonal floodplains, shrub woodland of riparian zones, tree shrubland with Acacia and shrub woodland with mixed mopane. A sandveld tongue stretches into NG 26 from the south-eastern boundary (Figure 2). The area is known to undergo significant changes between seasons, and from year to year as a result of flooding variations. The flood peak in this part of the Okavango Delta is usually between about April to August (Viljoen 2009). Figure 2. Abu fixed wing aerial survey strata determined by major land cover types 1. Thaoge R NG 8 Tribal Agriculture Conservation NG26 Wetland River Floodplains MOREMI GR Boro R NG 27a Dry woodland Seasonal floodplain Savannah Sandveld tongue ABU 3 Stratum ABU 2 Stratum Habu Stratum ABU 1 Stratum Fence NG 29 1 Google Earth Imagery (2007). 3
Methods Fixed-Wing Aerial Survey The survey used the standard methodology for strip transect sampling (Norton-Griffiths 1978), which has been well established for aerial surveys of large African herbivores (Chase & Griffin 2009, Craig & Gibson 2002). This report follows the procedures developed by Dunham et al. (2009) and Gasaway et al. (1986) for analyzing and presenting wildlife aerial survey data. Prior to the survey, NG 26 was subdivided into three strata, named Abu 1, Abu 2 and Abu 3. Systematic, parallel transects, were positioned in an east/west orientation across each stratum 1. The position of the first transect in each stratum was determined randomly using the DNR Garmin Sampling Extension (Minnesota Department of Natural Resources) in ArcView (ESRI, Redlands, CA, 2002). Transects were placed at right angles along generally east/west axes, corresponding to the perpendicular gradient of major watercourses and fence lines (Figure 2). These strata were largely delineated by changes in land cover, with each stratum becoming more inundated with water further east towards the Okavango Delta. Overall sampling intensity over the Abu survey area was planned to be 40 %, with a transect width (i.e. combined strip width of the two search strips) of about 400 m. The strata in NG 26, were expected to contain high densities of wildlife, and were sampled more intensively than the Habu stratum. Transect spacing therefore, varied from 1 km in the Abu concession to 2 km in the Habu stratum which is largely tribal agriculture land and which was expected to contain less wildlife (Figure 2). Transect lengths were typically < 15 km and could be flown in < 6 min, thereby reducing observer fatigue. Within the Habu stratum, planned sampling intensity was approximately 22 %. In the Habu tribal farming lands to the west of NG 26, the survey was intended to provide not only estimates of the numbers of large wild animals, but also information on the numbers and spatial distribution of human activities and cattle. The survey was designed using DNR Garmin software and ArcView (ESRI 2002). Stratum boundaries were mapped in ArcView, and the transect orientation and spacing was generated using the DNR Garmin Sampling Extension in ArcView (ESRI 2002). This software generates flight lines (transects), with the first flight line offset from the end of the stratum by an entered random number. The start and end points for each transects were transferred as waypoints to Mapsource (Mapsource 2007). Prior to flying, all transects were incorporated into a digital map of the survey area with their beginning and end point coordinates (Decimal Degrees, WGS 84). This digital map was created using ArcView 3.2 (ESRI, Redlands, CA, 2002) software and showed observable landmarks and boundaries. All transects were then mapped as routes (Mapsource 2007) prior to flying and then uploaded on a digital map within each GPS receiver (Tracks for Africa 2010) with their beginning and end point coordinates. 1 Stratum - is one block of a sample area which has been partitioned into blocks/strata (plural = strata). The results from the strata are then aggregated to make inferences about the population. 4
Flight Procedures The aerial survey was flown over four days, during the period 18 to 21 October 2010. The survey was conducted during the height of the hot-dry season when we expected increased visibility of wildlife (no rain had fallen during October). Transects were flown at an average speed of ~ 160 km per hour, at about 300 feet above ground level. Height above ground level was maintained using a Bendix King radar altimeter. The Plane was equipped with three GPS receivers. The pilot used one GPS to navigate along transect lines, a second GPS was used by the front seat data recorder to ensure the pilot did not deviate off the transect line, and the third GPS was used to record animal observations. The track log of the aircraft was recorded using the track log function in the GPS which noted the aircrafts location at intervals of 15 seconds (of time). Height, above ground level as indicated by the radar altimeter was recorded at regular intervals [~ between 20 to 40 seconds (of time)] along transects, regardless of whether or not animals were seen. Later the mean height for each transect was calculated. The time at which the flight along each transect was started and ended was also recorded to provide a guide to the average speed (Craig & Gibson 2002, Dunham et al. 2009). The aircraft crew included a pilot (Mike Holding), a data recorder (Michael Chase), who sat next to the pilot, and two observers (Kelly Landen and Adrian Dandridge) who sat behind the pilot and recorder. All four crew members were able to communicate efficiently through a four-way intercom headset receiver box. Observers in the Plane called out their observations into a recorder which had a 4 GB memory card. At the end of each day, recordings were played back on a laptop computer to verify hand-recorded data. The same two observers were used throughout the survey, one on each side of the plane. Both of the observers had extensive previous aerial survey experience (Kelly Landen > 1000 hrs, Adrian Dandridge > 50 hrs) prior to this project. For all strata we used the standard methodology for transect sampling developed by Norton-Griffiths (1978). Observers recorded wildlife species inside the counting interval when they were as nearly perpendicular to the plane as possible. Additionally, a mark was put on the plane window to help observers keep their eyes at a consistent height to maintain the same sighting angle for each observation (Appendix One). This helped to keep consistent interval widths for each observation. Any animals outside of the area delineated by these wands were not counted. For each observation seen within the transect interval, the observer called out the numbers of species. All animals seen by the observers within the search strips (see section Strip Width and Calibration below) were called to the recorder, who marked each observation as a waypoint in the GPS. The recorder also kept a written data log entered on a datasheet for each observation within the strip including: the waypoint number and time, altitude from the radar altimeter, number of individuals observed, and which observer made the sighting. The start and end times for each transect were also recorded to the nearest second. Transects lengths were provided by the using the animal movements extension in ArcView 3.2. On two occasions the aircraft left a transect that routed over an airstrip to provide the survey crew with a rest. The pilot and recorder used their GPSs to navigate to the location where the transect was stopped. 5
Observations During the survey observers were instructed to search for elephants, (and other large herbivores) buffalo, giraffe, hippo, impala, kudu, lechwe, tsessebe, wildebeest and zebra. In addition the following large birds, were counted: ostrich, bateleur eagle, fish eagle, open bill storks, saddle bill stork, and wattled crane. In the Habu stratum, cattle were counted. Elephants were recorded as being in family groups or bull groups. Family groups were herds, in which females and young were present, although the herd may have included elephant bulls. Bull groups were classified as single bulls or herds which contained no females or juveniles. The observers also recorded any elephant carcasses seen. All elephant carcasses noted were classified using four age categories as follows: Carcass category Definition 1 Fresh Carcass still had flesh, giving the body a rounded appearance. Vultures were probably present, and the ground still moist. (Likely to have died within the past month). 2 Recent Rot patch and skin still present. Skeleton not scattered. (Likely to have died within the past year). 3 Old Clean bones; skin usually absent; vegetation re-grown in rot patch. (Likely to have died more than 1 year ago). 4 Very Old Bones scattered and turning grey. (Likely to have died within the last 10 years). These carcass categories are those used by Douglas Hamilton & Hillman (1981), and recommended by the CITES programme Monitoring the Illegal Killing of Elephants (MIKE). Where possible, observers also noted the presence and absence of tusks from carcasses. To help determine if carcasses were possibly illegally hunted observers noted if tusks had been chopped or removed. This was done mainly when more than one elephant carcass was observed and in close proximity to each other so not to confuse potentially poached elephants with those hunted legally by sport hunters. Artificial waterholes, lodges and airstrips were identified. Those transects that ended over the Southern Buffalo Fence provided the recorder with an opportunity to note breaks in the Fence and make a general assessment of its structural integrity. In the Okavango Delta, there has been increasing concern about the impact elephants and fire are having on large trees e.g. baobab and the regeneration of other vulnerable tree species (Chase 2010). To provide current information on the status of baobab trees in the concession, observers were requested to count these trees, as well as assess the level of damage on each tree seen (assumed to have been caused by elephant). This damage/impact, was expressed as a percentage of the tree which had been eaten or impacted. The proportion of damage to a tree was categorised into one of five subjective categories: nil (0-10 %), light (11-6
30 %), moderate (31-50 %), heavy (> 50 %), and dead (100 %) (Knight et al. 1994). Baobab trees were also classified into the following three size categories: Baobab size Small Medium Large Definition < 1 m in diameter between 2-3 m in diameter. > 3 m in diameter. All wildlife species seen during the survey were recorded, although estimates (and assessments) of some were likely to be either inaccurate or imprecise. Estimates of small or cryptic species and those whose behaviour or habitat makes them difficult to see from the air can be inaccurate. Examples of these include kudu, lechwe, reedbuck and impala, among others. Rare species or those that have clumped distributions (such as sitatunga, buffalo and zebra, respectively) tend to have imprecise estimates (Craig & Gibson 2002). We tried to address these concerns by stratification of survey effort and aerial photography, which are two technical tools frequently used to improve precision and accuracy of wildlife surveys, respectively. Strip Width and Calibration Strip widths were delineated by two parallel aluminum wands connected to custom made brackets which were attached to each wing strut of the aircraft. The wands could be moved in any direction during the setup phase to delineate a planned 200 m parallel interval for recording wildlife observations at an altitude of ~ 91.5 m. Two digital EOS 7D Canon cameras (18 mega pixel) were fixed to specially adapted mounts on each window of the plane. The center of the lenses corresponded to marks on the plane window that were used to help observers keep their eyes at a consistent height for each observation. The focal lens was adjusted to incorporate the counting strip width, providing a consistent viewing angle. If any animal group was too large for all the individuals within it to be counted, a digital photo was taken. These photos were used to verify herd size and the sighting of herds within the interval defined by the wands. The components of the camera system consisted of two cameras with 20-mm wide-angle lenses, camera backs with time code generators, remote switches and two window camera mounts. A camera was mounted on each side of the plane. The cameras provided high-resolution photos so that animals could be more accurately counted during subsequent analyses. A GPS time code and date were recorded to the second for every frame exposed (Appendix One). Interval widths on each side of the plane were calibrated and confirmed prior to initiating a survey over each stratum. This was done by placing markers at measured distances (10 m apart) on the Abu airstrip and conducting flyover tests. After repeated flyovers (at ~ 91.5 m) and photo verification, wands were adjusted to provide a designated field of view for each strip interval of ~ 200 m at appropriate flight altitude. The aluminum wands were semi-permanently 7
attached to the struts for the duration of the survey. We flew transects during the morning (~ 0730 - ~ 1100 hrs) and afternoon (~ 1600 - ~ 1730) hours. Data Analysis Strip Transect Sampling / Fixed-Wing Aerial Survey Population estimates for individual strata were calculated following the guidelines developed by Norton-Griffiths (1978). We adjusted for altitude and photo corrections following Norton-Griffiths (1978) and used the traditional Jolly s Method II for unequal sized sampling units (Jolly 1969) to calculate population estimates and variance for each species in each stratum (Appendix 1). Variance estimates for strip transect counts were calculated from observation data collected within the mean combined strip width. The Jolly s Method II ratio method is based on the calculation of the ratio between animals counted and area searched. The population estimate is based on the density of animals per sample unit (transect) rather than number of animals per sample unit. Entire Survey Area and Strata within it. We calculated population estimates for each stratum and summed these estimates to obtain an estimate for our entire survey area. The upper and lower 95% confidence limits for population estimates for the entire survey area or stratum (within it) were calculated as: Where: Population estimate ±[t v x Square root of (Sum of Variances for individual strata)] V = the degrees of freedom estimated by Satterthwaite s rule (Gasaway et al. 1986) v was an integer, calculated using the formula: v = (Sum of Variances for individual strata) 2 Sum of [Variance for individual stratum) 2 / (n-1)] Comparison of observers. For each of the commoner species, the total numbers of individuals and groups counted by each observer in all transects was determined. For each observer and each species, the numbers of individual animals and groups that the observer was expected see was calculated. For each species, the observed and expected numbers of animals/groups seen were compared using Chi-square (X 2 ) one sample statistical tests with 1 degree of freedom (Dunham et al. 2009). Significant differences are reported at P < 0.05. Elephant carcasses. Following the method developed by Douglas Hamilton & Burrill (1991), and adapted by Dunham et al. (2009), the elephant carcass ratio (which is a percentage), defined as the ratio of dead elephant (of all categories) to all elephants (dead plus live animals), was calculated. 8
It is reasonable to assume that all category 1 or 2 carcasses represent elephant that may have died during 2010 (within the last year) Dunham et al. (2009). Hence, the 1+2 carcass ratio provides an index of elephant mortality (both natural and anthropogenic) during 2010 and was calculated as the estimated number of elephant carcasses in age categories 1 or 2 as a percentage of the sum of the estimated number of live elephants and the estimated number of carcasses in age categories 1 or 2. Photo-interpretation. High resolution digital photographs taken from cameras mounted on each side of the plane were used verify the numbers of animals seen by observers to those captured in the photos. This photo interpretation was especially helpful in counting large herds that are difficult to count from the air. In addition, photos helped to verify whether animals occurred within the counting interval (Norton-Griffiths 1978). Data analysis. Herd group sizes and distribution maps were created using ArcView, DNR Garmin software and Google Earth imagery. Photographs were viewed in Adobe Photoshop, colour corrected and dots placed on each counted animal within each counting group. Differences between observer and photo corrected counts were compared using Excel. Following the guidelines of Norton-Griffiths (1978), we tested for differences (d) between population estimates for the two surveys by comparing the sample errors for each estimate. Two sample t-tests were used to compare sizes of baobab trees seen between Abu and Habu. SYSTAT 10.2 and Excel were used for all statistical analyses. Search Effort The greater the time spent searching each square kilometre of a transect, the greater the probability that the observer saw animals that occurred within the counting strip. Search effort (in minutes per sq km) for a stratum was defined as the total time spent flying all transects within that stratum, divided by the total area of those same transects (Gasaway et al. 1986). Aerial surveys inherently underestimate wildlife numbers, with the degree of underestimation greater for small or cryptic species than for larger species. However, population estimates are given for all species, because the estimates provide useful indices of abundance (with measures of precision) that can be used to determine spatial distribution, as well as temporal trends in population number (Dunham et al. 2009). Other than the observations which were corrected by reference to photographs no other corrections have been applied to any estimates to compensate for any undercounting or missed animals. 9
Results Sampling Effort For the 2388 km 2 survey area, 162 transects (range 2-17 km), totaling 1996 km were flown over 12 hours (Table 1 & Figure 2). Flight altitude averaged 96.5 m (range 83-133 m) for wildlife observations. The search rate, (km 2 /min) was calculated as the total sample area divided by the total time on transects and averaged 1.2 for the entire survey area (Table 1). Table 1. Sampling statistics for the aerial survey of wildlife in Abu concession, Oct10. Stratum name Stratum area (km 2 ) Transect spacing (km) Number of transects [=n] Total transect length (km) Area covered (km 2 ) % of stratum sampled Abu 1 646 1 51 589 260 40% Abu 2 690 1 45 642 282 41% Abu 3 517 1 38 475 209 40% Subtotal / mean 1853 134 1706 751 41% Habu 535 2 28 290 112 21% Total / mean 2388 162 1996 863 28% Stratum name Time & date sampled Trans ect Flying time (hours) a Stratum Total Search effort (km 2 /min) Abu 1 am & pm 18 Oct 3.5 4.1 4.6 1.3 Abu 2 am 19 Oct 3.9 4.4 4.7 1.2 Abu 3 am 20 Oct 3.1 3.6 4 1.1 Subtotal / mean 10.5 12.1 13.3 1.2 Habu pm 21 Oct 1.5 1.9 2.3 1.2 Total 12 14 15.60 1.2 b a Transect time is the time spent searching the transects; stratum time is the transect time, plus the time spent flying between transects in the same stratum; and total time is the stratum time, plus the time spent traveling between the stratum and airstrip. b A search rate of 1.5 is considered adequate for large animals, but 1 or less should be aimed at for other animals (Gasaway et al. 1986, Craig & Gibson 2002). 10
Figure 3. Recorded track logs of flight paths during the Abu fixed-wing aerial survey, Oct10. NG 25 MOREMI GR Thaoge R Boro R NG 8 NG 27a ABU 3 Stratum ABU 2 Stratum Habu Stratum ABU 1 Stratum Survey Area Stratum Flight Path Fence NG 29 Wildlife Estimates The estimated numbers of elephants, elephant bulls, elephant family groups, elephant carcasses (age category 3), buffalo, giraffe, hippopotamus, impala, kudu, lechwe, ostrich, reedbuck, tsessebe, wildebeest and zebra, baobab, birds and cattle are given in Tables 8 to 24 respectively. Estimates are given for each stratum, and for the entire Abu concession. There may appear to be small arithmetic errors in some tables, but these are rounding errors. If the number of individuals seen (No. seen) is greater than the lower confidence limit (Lower CL), then it is biologically meaningful to replace the calculated lower confidence limit with the number seen (Chase & Griffin 2009, Dunham et al. 2009). For practical purposes, it can be assumed that the number of a given species in a given stratum lies between the lower and upper confidence limits, with the estimate providing the best estimate of the number there. For example, from Table 2, one can say that there were between 3042 and 4169 elephant in Abu concession, with 3606 being the best estimate of the number of elephants in the concession. For practical purposes, one might say that there were 11
between 3000 and 4200 elephants in Abu during the late dry season of 2010, with 3600 being the best estimate of the number of elephants there. Low numbers of waterbuck, roan (two of each seen in Abu 1) gemsbok (three in Abu 3), lion (one in Abu 2), and sitatunga (one in Abu 3) were seen during the survey, and no attempt has been made to estimate the numbers of these species. While baboons were seen their numbers have not been estimated. Table 2. Wildlife estimates and statistics for major wildlife species, elephant carcasses, baobab trees and birds in Abu 1, NG 26 WMA during aerial survey, Oct10. Species Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) Woodland Sp Elephant 3606 1453 81527 16 3042 4169 1.95 Elephant Bulls 531 213 2364 18 435 627 0.29 Elephant Family 3075 1240 76238 18 2528 3621 1.66 Ele Carcass 3 65 25 104 31 44 85 0.04 Buffalo 3745 1523 740163 45 2043 5446 2.02 Giraffe 426 173 3980 29 310 550 0.23 Impala 2705 1094 176939 31 1873 3537 1.46 Kudu 308 125 2051 29 218 397 0.17 Tsessebe 278 111 2352 34 182 374 0.15 Warthog 131 51 427 31 90 171 0.07 Wildebeest 151 57 2407 64 57 248 0.08 Zebra 1303 524 57010 36 831 1775 0.70 Wetland Sp Hippo 287 117 3055 38 177 396 0.15 Reedbuck 58 24 156 41 33 83 0.03 Lechwe 1995 811 47744 22 1563 2427 1.08 Bird Sp Bateleur eagle 94 47 258 66 62 126 0.05 Ostrich 97 35 1058 66 35 161 0.05 Saddle-bill stork 154 61 449 27 112 196 0.08 Wattled crane 148 59 767 36 93 202 0.08 Fish eagle 146 59 524 31 101 191 0.08 1 Estimates for observations seen in Habu are provided in Tables 8 to 24. Densities and estimates from Abu strata 1-3 (Tables 8 to 24) are shown separately from the data of Habu stratum because they were surveyed at different sampling intensities, i.e. ~ 40 % versus ~ 20 % coverage. 12
Observations During the survey observers recorded 1379 herd observations, with stratum Abu 1 having the most observations (429) (Table 3). Elephant (247), lechwe (156), Impala (100), and zebra (78) herds were observed frequently. Buffalo had the largest average herd size (32.5), although one herd numbered ~167 buffalo. Average herd sizes for elephant family groups and zebra were similar (12). The most numbers seen occurred in the Habu stratum, and can be attributed to the high number of cattle counted in this area. Abu 1 stratum had the most herd observations, and would have had the highest wildlife had it not been for the openbill storks counted in Abu 2 stratum. Baobab trees (163) mostly occurred in the Abu 1 stratum as did elephant carcasses of all categories (24). Birds of national importance included wattled crane (57), saddle-bill stork (83) and fish eagle (59), which were largely seen in Abu 2 and 3 strata. Table 3. Numbers seen, herds observed and average herd size of animals, baobabs and carcasses seen in each stratum. Species No seen Total No. Herd Obs Total Av. Herd Size Total Abu 1 Abu 2 Abu 3 Habu Abu 1 Abu 2 Abu 3 Habu Abu 1 Abu 2 Abu 3 Habu Elephant bull 137 46 30 15 228 77 29 18 9 133 1.8 1.6 1.7 1.7 1.7 Elephant cow 489 438 313 20 1260 41 40 32 1 114 11.9 13.3 10.5 20 12.1 Buffalo 294 450 779 3 1526 18 10 17 1 46 16.3 45 46 3 32.5 Baboon 88 20 40 6 154 6 1 3 1 11 14.7 20 15 6 13.4 Baobab 65 37 14 47 163 51 36 14 15 116 1.3 1 1 3.1 1.4 Bateleur 6 8 3 17 6 8 3 17 1 1 1 1 Cattle 2 3657 3659 1 193 194 2 19 19 Ele carcass 1 2 1 3 2 1 3 1 1 1 Ele carcass 3 22 2 1 1 26 21 2 1 1 25 1 1 1 1 1 Fish eagle 5 15 39 59 5 14 31 50 1 1.1 1.2 Gemsbok 3 3 1 1 3.0 3 Ground hornbill 8 7 15 3 2 5 3.5 3 Giraffe 61 87 25 173 26 26 7 59 2.4 3.3 3.6 3 Hippo 15 42 60 16 133 3 9 14 3 29 5.0 4.6 4.3 5.3 4.6 Impala 625 328 141 21 1115 52 28 18 2 100 15.6 11.7 7.8 10.5 13 Kudu 61 49 15 9 134 14 15 8 3 40 4.4 3.3 2 3 3.3 Lechwe 49 177 585 2 813 6 40 109 1 156 8.2 4.5 5.4 2 5.2 Lion 1 1 1 1 1 Openbill stork 900 3888 4788 4 7 11 225 555 435 Ostrich 30 3 2 55 90 8 2 1 22 33 3.8 1.5 2 2.5 2.7 Reedbuck 9 8 7 2 26 6 5 6 2 19 1.5 1.6 1.2 1 1.4 Roan 2 2 1 1 2.0 2 Saddle-billed S 9 17 35 2 63 7 11 20 2 40 1.3 3.4 1.8 1 2 Sitatunga 1 1 1 1 1.0 1 Tsessebe 41 42 28 111 10 10 9 29 4.1 4.2 3.1 3.8 Warthog 30 16 5 51 16 9 3 28 1.9 1.8 1.7 1.8 Waterbuck 2 2 1 1 2 2 Wattled crane 10 14 35 59 3 7 17 27 2.7 2.0 2.1 2.1 Wildebeest 30 18 9 57 7 2 2 11 4.3 9.0 4.5 5 Zebra 413 66 45 441 965 35 13 6 24 78 11.8 5.0 7.5 18.4 12.4 Total /mean 3403 5772 2224 4298 15697 429 325 344 281 1379 13.8 31.6 5.5 6.2 19.7 13
Comparison of observers. A comparison of the numbers of animals seen by the two observers (Table 4), suggested that generally they saw similar numbers of herds of animals. The left observer counted more individual animals. It is not possible from the survey results to determine which observer counted most accurately. No Chi-square test was conducted if any expected number was < 5. Table 4. Comparison of numbers of individual animals seen and numbers of herds/groups seen by the left and right observers. Species Obs. No of Expected No of Obs. No of herds Expected No of Chi-square individuals P Chi-square herds individuals individuals herds Left Right Left Right Left Right Left Right Elephant bull 126 102 113 115 77 56 66 67 3.2 0.076 3.9 0.050 Elephant family 755 622 680 697 66 48 56 59 16.3 0.000 3.7 0.055 Buffalo 469 1060 755 774 28 19 23 24 720 0.000 2.2 0.140 Baboon 119 96 106 109 9 4 6 7 3.1 0.080 2.1 0.150 Baobab 100 61 80 81 70 44 56 58 9.9 0.002 6.8 0.009 Bateleur 6 11 8 9 6 11 8 9 0.9 0.331 1.3 0.259 Cattle 1953 1706 1807 1852 92 102 96 101 23.4 0.000 0.2 0.690 Ele carcass 3 15 11 13 13 14 11 12 13 0.7 0.397 0.6 0.449 Fish eagle 32 27 29 30 28 22 25 26 0.6 0.455 1.2 0.276 Ground hornbill 7 8 7 8 2 3 2 3 0.0 0.834 0.1 0.718 Giraffe 106 67 85 88 37 22 29 32 9.8 0.002 5.1 0.025 Hippo 90 43 66 67 17 12 14 16 17.8 0.000 1.3 0.247 Impala 869 430 641 658 58 42 49 54 0.0 0.000 4.2 0.040 Kudu 62 75 68 69 21 20 20 22 0.9 0.334 0.3 0.609 Lechwe 429 384 401 412 84 71 77 85 3.7 0.053 2.9 0.088 Ostrich 48 42 44 46 21 12 16 18 0.6 0.453 3.5 0.061 Reedbuck 17 9 13 13 12 7 9 11 2.7 0.103 1.9 0.165 Saddle-billed S 58 25 41 42 23 17 20 22 14.0 0.000 1.8 0.174 Tsessebe 76 35 55 56 16 13 14 16 16.2 0.000 0.9 0.342 Warthog 25 26 25 26 16 12 14 16 0.0 0.959 1.3 0.254 Wattled crane 33 24 28 29 17 10 13 15 1.7 0.198 2.9 0.088 Wildebeest 39 18 28 29 7 4 5 6 8.3 0.004 1.3 0.252 Zebra 674 301 481 494 47 32 39 46 442.8 0.000 5.7 0.017 P The left observer saw more individual animals (except for buffalo) than the right observer. The left observer also saw more herds of wildlife (for all species) than the right observer (Table 4), although for many of these observations the two observers saw approximately similar numbers of herds. Some species (e.g. buffalo, impala and zebra) often occur in relatively large herds. Furthermore, relatively few groups of these same species are seen during the survey. Hence, some of the differences, while statistically significant, are probably a consequence of chance. However, differences were found for most species, the left observer saw more animals than expected if the observers were of similar efficiency. Comparison of observers and photographs. We used 112 photographs to verify the numbers of animals counted by the observers. Observers were accurate in estimating animal numbers in herd sizes < 40, greater than this number observers tended to underestimate animals (Table 5). Only photos with > 40 animals were used to compare observer estimates. The left observer s propensity to take more photos (69) which were used in subsequent analysis may account for the differences in wildlife numbers seen between the two observers. 14
Table 5. Comparison between observer estimates and photographs of the same animals. Species Observer numbers Photo numbers Av. Herd size % difference Elephant bull 222 229 1.72 3.2% Elephant family 1272 1377 12.1 8.3% Buffalo 880 1529 33 73.8% Cattle 3649 3659 17 0.3% Giraffe 173 173 3 0.0% Hippo 133 133 4.6 0.0% Impala 1225 1311 13 7.0% Lechwe 807 818 5.2 1.4% Open-billed S 3400 4788 435 40.8% Ostrich 90 90 2.7 0.0% Tsessebe 113 111 3.8-1.8% Warthog 51 53 1.8 3.9% Wattled crane 57 59 1 3.5% Wildebeest 55 59 5 7.3% Zebra 909 995 12.4 9.5% Grand Total 13036 15384 18.0% Elephant carcasses. Just two recent elephant carcasses (i.e. age category 1) were seen during the survey. The majority of elephant carcasses seen (26) occurred in the age category 3. Douglas-Hamilton et al (1991) suggest a carcass ratio of 2 8 % as being normal for a stable or increasing population. The estimated total number of elephant carcasses (70) in the survey area during Oct10 represented 1.4 % of the estimated total number of live and dead elephants. The estimate for recent carcasses produced an estimate of 5 fresh elephant carcasses in the Abu 1 stratum. The recent category 1 carcass ratio (which reflects the mortality rate of elephant during the survey year) was 0.14 %. Baobab observations. A total of 163 baobab trees were seen during the survey, of which 65 occurred in the Abu 1 stratum. The majority of baobab trees seen were in the medium size category (76). Dead trees (100 % damage) accounted for 6 % of the total number seen within the Abu concession. Small size baobabs make up 17 % of all baobabs which occurred in the survey area (Table 6). The majority of small baobabs (83 %) occurred west of the Southern Buffalo Fence in the Habu tribal lands where elephant densities are currently low (0.32 km 2 ). The number of small baobabs seen in the Abu and Habu concession differed (t = 2.6, df = 9, P = 0.02), with Habu having 68 % more small trees (n = 37) than Abu (n = 22). 15
Table 6. Number, size and the extent of damage to baobab trees seen during the aerial survey. Stratum % of Damage No seen Size No seen Abu1 100 2 large 19 0-10 51 medium 37 11-30 12 small 9 subtotal 65 2.4 Abu2 100 2 large 13 0-10 22 medium 12 11-30 11 small 12 31-50 2 subtotal 37 1.7 Abu3 100 3 large 4 0-10 10 medium 9 31-50 1 small 1 subtotal 14 1.6 Habu 0-10 45 medium 10 11-30 2 small 37 subtotal 47 0.24 Total 163 163 Elephant density (km 2 ) Wildlife Distributions The spatial distribution of wildlife is illustrated in Figures 6 to 22. On most maps, the distribution is shown as the location of sightings of groups of the given species, together with an indication of the size of group/herd. Wildlife was widely distributed throughout the concession, this can be attributed to both the diverse number of species found in the concession and the heterogeneous habitats which occupy the concession (Figure 3). Elephants were seen in most parts of the survey area, but preferred the drier western areas of the concession (Figure 6). Buffalo observations mainly occurred along the rivers and adjoining floodplains (Figure 10). Giraffe were mostly observed in dry savanna habitat in the southwest of the concession (Figure 11). Hippo occurred along the main river channels, floodplains and lediba s (lagoons) in the northeast and near Abu camp (Figure 12). Impala occurred throughout the dry savanna plains and woodlands which dominate the southeast corner of the concession (Figure 13). Lechwe were largely distributed adjacent to the Boro River in the northeast (Figure 15). Kudu and tsessebe, occurred largely throughout the concession (Figure 16). Wildebeest and zebra were observed on the dry savanna grasslands, primarily in the western side of the concession (Figure 19 & 20). 16
The distribution of birds of which many are small or cryptic species that are hard to survey from the air or that had < 10 observations are shown in Figure 21. Even though these species are not easily surveyed from the air, this map is included for possible comparison with the distribution maps from the earlier survey (Viljoen 2009) and future surveys. Figure 3. The distribution of all wildlife seen during the aerial survey of Abu (NG 26), Oct10. Thaoge R Boro R MOREMI GR NG 27a NG 8 Potae R NG 26 Fence Cattle Wildlife NG 29 Habu (NG8) Survey The Habu community lands (NG 8) are pastoral and arable cattle grazing areas on tribal land and share the western border with Abu. This community area is separated from Abu WMA by the Southern Buffalo Fence. During our survey of Habu we recorded the spatial distribution of cattle (Figure 4 & Figure 24); this also provided us with an opportunity to assess the integrity of the Southern Buffalo Fence. Many sections of the fence were broken, and appeared to have been breached by elephants. We observed no fence maintenance crews attempting to fix down sections of the fence. This task is made more difficult by the high flooding the area is experiencing, and which may be the reason for its current state. Cattle (15) were observed within the Abu concession when flying from the airstrip to the survey area. Considering the high density of cattle in Habu (32 / km 2 ), and the need for additional grazing, the fence acts as important barrier limiting the movement of cattle into Abu. 17
The Habu survey area is an important wildlife refuge, especially for the 2093 zebra we estimated in this farming area (Table 22). There is an opportunity for community based ecotourism which could provide a more viable and ecologically sustainable form of land use in this area. With increasing human pressure in this area, the future management of this large zebra population should be considered by government authorities, especially when zebra numbers appear to be declining in northern Botswana (Chase unpublished data). One option to conserve this population could be the realignment of the Southern Buffalo Fence, the creation of a wildlife corridor or the creation of a community conservation conservancy, further research is needed in this area. Figure 4. Distribution of cattle and wildlife in Abu (NG 26) and Habu (NG 8). NG 25 MOREMI GR Thaoge R NG 8 NG 26 NG 27a Fence NG 29 Possible Fence Realignment A comparison of wildlife density and distribution shows that there is a clear separation between areas of high cattle and high wildlife numbers (Figure 4). Wildlife density in the Habu area was 3 animals / km 2, while in the Abu concession is was 40 animals / km 2. Where people and livestock are concentrated, wildlife populations are lower. 18
Trends in Wildlife Numbers in Abu Two aerial surveys of wildlife have been conducted in the Abu concession, (Viljoen 2009 & Chase 2010). Other than the difference in sampling coverage, both surveys used similar methods. Differences in wildlife estimates/counts between these two surveys are presented in the table below. Table 7. Estimated numbers, densities and differences between estimated numbers of animals Sept09 and Oct10, aerial surveys. 2009 Survey Estimates 2010 Survey Estimates Species Estimate No. Density Estimate No. Density seen (km 2 ) seen (km 2 ) % pa Change 2009-2010 d trend Elephant 1251 1251 a 0.68 3606 1453 2.11 65 8.18 Buffalo 2562 2562 1.38 3745 1523 2.21 32 1.37 Giraffe 328 73 0.18 426 173 0.25 23 1.14 Impala 3092 707 1.67 2705 1094 1.59-14 0.7 Kudu 573 131 0.31 308 125 0.15-86 6.3 Tsessebe 307 71 0.17 278 111 0.16-10 0.51 Zebra 901 208 0.49 1303 524 0.77 31 1.61 Lechwe 1618 370 0.87 1995 811 1.17 19 1.48 a Elephant and buffalo on the Sept09 survey (Viljoen 2009) were counted both in and outside the strip interval. % = percent change between 2009 and 2010. d = value for comparison between years. = likely upward trend; = upward trend or stable; = downward trend or stable; = probably stable population. The estimation of population trends from a small number of surveys is subject to changes in conditions masking the real trends (Craig & Gibson 2002). Comparing two surveys is the simplest way of detecting a trend. The numbers of elephants seen on the two aerial surveys was similar; however the estimate of this survey is significantly higher (d > 1.96 for all d tests) than the Sept09 survey (Table 7). Both surveys yielded similar estimates for impala and tsessebe. The Oct10 survey yielded significantly lower estimates for kudu (d = 6.3). The last column in Table 7 indicates the likely direction of trend. In the case of elephants, the prediction of an upward trend takes into account the low death rate inferred from recent carcass numbers. 19
Discussion Wildlife Estimates The managers of the concession (together with the survey consultant) felt it pertinent that this survey increase the precision of population estimates for Abu concession (i.e. to reduce the confidence interval of the estimate, but the population estimate itself may be biased, that is, usually on the low side). Therefore, this survey was conducted in a more robust manner, at high sampling coverage (~ 40 %), which is appropriate for Abu where large herbivores are numerous. It was felt that a high sample aerial survey would yield accurate estimates of wildlife in the concession and establish a sound basis for wildlife numbers needed to follow population trends. These data can now be used to evaluate the efficacy of conservation activities and trends in wildlife populations with the ability to prioritize conservation projects against any severe variations in wildlife numbers. Observations Comparison of observers. Photos provide an important tool for obtaining more accurate herd counts. Both our aerial observers had experience in counting animals from the air, which may account for why this survey estimated more elephants and buffalo in the concession than the Viljoen (2009) aerial survey. The left observer did see more animals than the right observer; this may be attributed to the left observer taking 35 % more photos which were used to verify herd sizes. Elephant carcasses. Many of the category 3 elephant carcasses seen during the survey occurred in the Abu 1 stratum which was previously used for professional sport hunting. This activity was stopped in 2008. The two recent elephant carcasses we observed still had their tusks intact, and suggest a natural cause of death. The estimated elephant mortality rate (0.14 %) is currently low in the concession. Baobabs. The Abu concession has already built protective barriers around eight baobabs near their lodges. This survey suggests that the current mortality rate of baobabs in the concession is between 5-6 %, (assuming the 100 % destroyed baobabs observed during the survey were killed within one year). Given the high density of elephants within the concession (2 / km 2 ), and the low number of small baobabs, managers are encouraged to continue with this tree protection programme and to extend it to include other tree species. In addition, and as a matter of priority, natural resources managers in the concession should establish a seedling project, to encourage the regeneration of indigenous trees of concern in the Delta (Chase 2010). Incidental observations made by the survey crew noted that woodlands were less impacted by elephant browsing west of the fence in the Habu area (0.32 elephant / km 2 ). The fence has acted as an effective barrier, limiting the movements of elephants west into this farming area resulting in reduced impact on the vegetation. 20
Wildlife Distribution The absence of long term aerial survey data for the Abu concession makes it difficult to infer any solid conclusions about wildlife distribution. The distribution of wildlife is likely to vary according to seasonal climatic conditions, the timing of the flood and the movements of animals. Wildlife was widely distributed throughout the concession, this can be attributed to both the diverse number of wildlife species and the heterogeneous habitats which occur within the concession. The distribution and density of elephants (in both Abu and Habu) provide a valuable parameter to monitor the pressure of use on the habitat. The northeast corner of the concession had numerous wildlife observations. Trends in Wildlife Numbers Population of most species in the area are probably increasing (if changing at all), only low rates of increase are biologically possible and low rates are not currently detectable with the existing precision. Prior to 2008, Abu concession allowed hunting in the west (Abu 1) portion of the concession. While DWNP have conducted aerial surveys over the concession, their survey stratum were not delineated along Abu s concession boundary, rather their survey over this area included other concessions. Viljoen s (2009) aerial survey is the only other survey conducted over the concession. Without any long term or reliable data for wildlife estimates, it is difficult to asses the trend in wildlife numbers. An aerial survey is a snapshot in time of the abundance and distribution of animals in a defined sample area. However, based upon the results of our survey it can be assumed that wildlife numbers have increased, this increase can be attributed to the following: 1. The cessation of hunting in the concession may have provided wildlife with the requisite environment conducive for their numbers to increase, but more importantly, created an opportunity for wildlife to disperse into the area; 2. The distribution of wildlife on open unfenced concessions adjacent to Moremi GR and other WMA, will continue to provide variable estimates on the distribution and abundance of wildlife. Wildlife, will continue to move from one bank of the river to the other, disperse to newly flooded regions which are not covered by these aerial surveys or move across park boundaries (Chase 2007, Martin 2005), all leading to variability in wildlife numbers. Chase (2007) has noted that elephants which spend the dry season in the Abu concession, move 400 km east across the Okavango into the Mababe Depression; and, 3. Species possibly clumped together, as a large portion of the survey area (and other parts of the Delta) are under water (Worski 2010). Many species have increased faster than is possible through natural population growth and this suggests that there have been variations is wildlife movements, dispersal, improvements in sighting conditions, aerial survey methods and availability of water. The two aerial surveys conducted over Abu provide seminal baseline information from which future surveys can infer population trends, and a reliable monitoring programme put into place. Conducting aerial surveys in a standardized way should lead to repeatable and comparable indices of abundance. Having started out with a broad scale sampling design (possible under 21
sampling Viljoen 2009), this survey has provided valuable baseline information about wildlife numbers. Recommendations Any time-series analysis depends on the use of similar (ideally identical) methods during successive surveys, so that any observed differences in population number can confidently be assumed to be genuine and not simply a consequence of changing methods. Given the difficulty of ensuring that methods are identical (e.g. the same observers are often not available for successive surveys), the application of high and consistent standards during the execution of surveys is important. Future aerial surveys should standardize their coverage (20-30%). Effective conservation management requires a good understanding of wildlife population dynamics and reliable estimates of population densities. This is especially true in many of Botswana s concession areas, where ecotourism is the main source of income. To provide accurate estimates of population sizes of large animals ranging over extensive areas, aerial surveys are often the only practical way to monitor wildlife trends. I suggest that future surveys should be conducted, which would provide critical information on wildlife distribution, abundance and trends. It has been pointed out that changes in conditions between surveys may be affecting the comparability of surveys despite the standardisation of methods and observers. It would be impossible to standardize the vegetation cover, air clarity or the weather. However, one improvement might be to standardize the survey dates. Ideally, the managers of Abu need to conform to one time period in which to conduct these aerial surveys. Conducting surveys at different time periods makes it difficult to infer any reliable conclusions about trends given the temporal and spatial variability associated with wildlife movements. Surveys should not take place later than the third week in October. These surveys however still provide an opportunity to evaluate wildlife distribution, abundance, trends and to compare changes over time specific to when the surveys were conducted. Ground counts should be encouraged as these would provide more reliable estimates for those species which are difficult to count from the air and which we estimated lower numbers, (impala, lechwe, tsessebe and kudu). These counts would also be valuable in determining wildlife population trends. Conclusion This survey has emphasized the value of aerial surveys for monitoring wildlife trends in animal numbers, even though the variation in numbers may be biologically impossible. Apart from their immediate value in management, aerial surveys of wildlife conducted at the concession level, will often provide data which will greatly improve wildlife managers understanding of the population dynamics of several large herbivores species in the Okavango Delta. Further, information on population sizes of individual species can also be used to set priorities, allowing conservation effort to be focused on those species most in need of attention e.g. the zebra population in the Habu area. In addition, surveys can be used to collect 22
information on where wildlife are in relation to different habitats, and so assess habitat associations e.g. lechwe. Private concession level aerial surveys are critical when current DWNP aerial surveys are intermittent and do not cover the area at the sampling intensity required to detect trends at finer spatial scales. Indeed, the execution, data compilation and analysis of this survey, provides a good example of collaboration and mutualism between government management and private concession holders. The current aerial survey routine should now be continued. 23
References Chase, M.J. and C. Griffin. (2009) Elephants caught in the middle: Impacts of war, fences, and people on elephant distribution and abundance in the Caprivi Strip, Namibia. African Journal of Ecology 47, 223-233. Chase, M.J. (2010) Ecology of elephants in the Okavango Delta. Elephants Without Borders, Kasane, Botswana. Craig, C. and D. St.C. Gibson. (2002) Aerial survey of Wildlife in the Niassa Reserve and Hunting Concessions, Moçambique, October/November 2002. Dunham, K.M., E. van der Westhuizen, H.F. van der Westhuizen and E. Gandiwa. (2009) Aerial Survey of Elephants and other Large Herbivores in Gonarezhou National Park (Zimbabwe), Zinave National Park (Mozambique) and surrounds. Frankfurt Zoological Society. DNR Garmin Extension. (2002) Minnesota Department of Natural Resources, GIS Section. Available from: http://www.dnr.state.mn.us/mis/gis/tools/arcview/ extensions/dnrgarmin/dnrgarmin.html (accessed June 2002). Fennessy, J. (2009) The Final Gene: Establishing the genetic architecture of the northern Botswana giraffe population. Giraffe Conservation Foundation. Progress Report Botswana Government. Gaborone, Botswana. GARMIN Corporation. (2000) MapSource: User s manual and reference guide. GARMIN International, Olathe, Kansas, USA. Gasaway W.C., S.D. dubois, D.J. Reed and S.J. Harbo. (1986) Estimating moose population parameters from aerial surveys. Biological Papers of the University of Alaska No. 22. 108pp. Google Earth. (2007) Google. http://earth.google.com/. Jolly, G.M. (1969) Sampling methods for aerial census of wildlife populations. East African Agriculture and Forestry Journal 34, 46-49. Norton-Griffiths, M. (1978) Counting animals. African Wildlife Leadership Foundation. Nairobi, Kenya. pp. 139. Viljoen, P. (2009) Aerial survey of wildlife in the Abu (NG 26) Wildlife Management Area. Elephant Back Safaris. Maun. Worski, P. (2010) http://www.orc.ub.bw/datacat/thamalakane2010_april.php. Accessed 28 November 2010. 24
Figure 5. Land use map and key features in and surrounding Abu (NG 26) WMA. NG 25 MOREMI GR NG 8 Thaoge R Tribal cattle grazing / agriculture Abu NG 26 Boro R NG 27a NG 29 Southern Buffalo Fence Map Datum: WGS 84, Decimal Degrees 25
Table 8. Population estimates and statistics for elephants in Abu (NG 26) and Habu. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 1577 626 32657 23 1214 1940 2.44 Abu 2 1181 484 27253 28 848 1513 1.71 Abu 3 848 343 21617 35 550 1146 1.64 Subtotal 3606 1453 81527 16 3042 4169 1.95 HABU/ NG8 171 35 7114 99 35 340 0.32 Total 3777 1488 88641 16 3189 4364 1.58 Figure 6. Distribution of elephants in Abu and Habu, Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 27a NG 26 Fence NG 29 The points indicate the locations of elephant seen within the search strips, together with an indication of the size of each herd. The dark black line indicates the area sampled, (to identify other features refer to Figure 5). 26
Table 9. Population estimates and statistics for elephant bulls in Abu (NG 26) and Habu. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 345 137 1785 25 260 430 0.53 Abu 2 112 46 384 35 58 152 0.16 Abu 3 74 30 195 38 45 102 0.14 Subtotal 531 213 2364 18 435 627 0.29 HABU/ NG8 76 16 582 63 27 124 0.14 Total 607 229 2946 18 500 714 0.25 Figure 7. Distribution of elephant bulls in Abu (NG 26) and Habu, Oct10. NG 25 MOREMI GR Thaoge R Boro R NG 8 NG 26 NG 27a NG 29 Fence 27
Table 10. Population estimates and statistics for elephant family groups in Abu (NG 26) and Habu. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU/ NG26 Abu 1 1232 489 32041 30 865 1600 1.91 Abu 2 1069 438 23494 29 760 1377 1.55 Abu 3 774 313 20703 38 482 1065 1.50 Subtotal 3075 1240 76238 18 2528 3621 1.66 HABU/ NG8 94 20 7017 179 20 263 0.18 Total 3169 1260 83255 18 2599 3738 1.33 Figure 8. Distribution of elephant family groups in Abu (NG 26) and Habu, Oct10. NG 25 MOREMI GR Thaoge R Boro R NG 8 NG 26 NG 27a NG 29 Fence 28
Table 11. Estimates of elephant carcasses in Abu (NG 26) and Habu, Oct10. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 55 22 89 35 36 73 0.09 Abu 2 5 2 7 100 2 10 0.01 Abu 3 5 1 8 160 1 13 0.01 Subtotal 65 25 104 31 44 85 0.04 HABU/ NG8 5 1 10 150 1 9 0.01 Total 70 26 114 30 49 91 0.03 Figure 9. Distribution of elephant carcasses (category 3) in Abu (NG 26) and Habu, Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 26 NG 27a Fence NG 29 29
Table 12. Population estimates and statistics for buffalo in Abu (NG 26) and Habu. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 716 294 102716 90 294 1360 1.11 Abu 2 1098 450 206113 83 450 2013 1.59 Abu 3 1931 779 431334 69 779 3262 3.74 Subtotal 3745 1523 740163 45 2043 5446 2.02 HABU/ NG8 14 3 155 179 3 39 0.03 Total 3759 1526 740318 45 2060 5458 1.57 Figure 10. Distribution of buffalo herds in Abu (NG 26) and Habu, Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 26 NG 27a NG 29 Fence 30
Table 13. Population estimates and statistics for giraffe in Abu (NG 26). Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 152 61 1519 51 74 230 0.24 Abu 2 212 87 2084 43 120 304 0.31 Abu 3 62 25 377 63 25 101 0.12 Total 426 173 3980 29 310 550 0.23 Figure 11. Distribution of giraffe in Abu (NG 26), Oct10. NG 25 MOREMI GR Thaoge R NG 26 NG 8 NG 27a NG 29 Fence 31
Table 14. Population estimates and statistics for hippopotamus in Abu (NG 26) and Habu. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 37 15 373 105 15 76 0.06 Abu 2 102 42 1181 68 42 171 0.15 Abu 3 148 60 1501 53 70 227 0.29 Subtotal 287 117 3055 38 177 396 0.15 HABU / NG8 76 16 2068 122 16 169 0.14 Total 363 133 5123 39 222 504 0.15 Figure 12. Distribution of hippopotamus in Abu (NG 26) and Habu, Oct10. NG 25 MOREMI GR Thaoge R NG 26 NG 8 NG 27a NG 29 Fence 32
Table 15. Population estimates and statistics for impala Abu (NG 26) and Habu. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 1557 625 150790 50 777 2337 2.41 Abu 2 800 328 22525 38 498 1103 1.16 Abu 3 348 141 3624 35 226 470 0.67 Subtotal 2705 1094 176939 31 1873 3537 1.46 HABU / NG8 100 21 3778 126 21 225 0.19 Total 2805 1115 180717 30 1965 3644 1.17 Figure 13. Distribution of impala Abu (NG 26) and Habu, Oct10. NG 25 MOREMI GR Thaoge R NG 26 NG 8 NG 27a Fence NG 29 33
Table 16. Population estimates and statistics for kudu Abu (NG 26) and Habu. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 152 61 1231 46 81 222 0.24 Abu 2 119 49 685 45 67 172 0.17 Abu 3 37 15 135 62 15 61 0.07 Subtotal 308 125 2051 29 218 397 0.17 HABU / NG8 42 9 1160 167 9 112 0.08 Total 350 134 3211 32 238 462 0.15 Figure 14. Distribution of kudu Abu (NG 26) and Habu, Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 27a Fence NG 29 34
Table 17. Population estimates and statistics for lechwe Abu (NG 26), Oct10. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 122 49 7562 143 49 296 0.19 Abu 2 429 177 5869 36 275 584 0.62 Abu 3 1444 585 34313 23 1068 1819 2.79 Total 1995 811 47744 22 1563 2427 1.08 Figure 15. Distribution of lechwe Abu (NG 26), Oct10. MOREMI GR NG 8 NG 26 NG 27a Fence 35
Table 18. Population estimates and statistics for ostrich in Abu (NG 26) and Habu. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 74 30 1010 86 30 188 0.11 Abu 2 7 3 17 114 3 15 0.01 Abu 3 16 2 31 125 2 36 0.03 Subtotal 97 35 1058 66 35 161 0.05 HABU / NG8 261 55 6256 62 99 423 0.49 Total 358 90 7314 47 189 527 0.15 Figure 16. Distribution of ostrich in Abu (NG 26) and Habu, Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 26 NG 27a Fence NG 29 36
Table 19. Population estimates and statistics for reedbuck Abu (NG 26) and Habu, Oct10. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU/ NG26 Abu 1 22 9 68 73 9 40 0.03 Abu 2 19 8 60 79 8 35 0.03 Abu 3 17 7 28 59 7 28 0.03 Subtotal 58 24 156 41 33 83 0.03 HABU/ NG8 9 2 33 122 2 21 0.02 Total 67 26 189 40 40 94 0.03 Figure 17. Distribution of reedbuck Abu (NG 26) and Habu, Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 26 NG 27a Fence NG 29 37
Table 20. Population estimates and statistics for tsessebe Abu (NG 26), Oct10. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 102 41 856 57 43 161 0.16 Abu 2 107 42 1176 64 42 176 0.16 Abu 3 69 28 320 52 33 105 0.13 Total 278 111 2352 34 182 374 0.15 Figure 18. Distribution of tsessebe in Abu (NG 26), Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 27a NG 26 Fence NG 29 38
Table 21. Population estimates and statistics for wildebeest Abu (NG 26), Oct10. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 80 30 1026 80 30 144 0.12 Abu 2 44 18 1020 145 18 108 0.06 Abu 3 27 9 361 141 11 65 0.05 Total 151 57 2407 64 57 248 0.08 Figure 19. Distribution of wildebeest Abu (NG 26), Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 27a Fence NG 29 39
Table 22. Population estimates and statistics for zebra Abu (NG 26), and Habu. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 1031 413 54089 45 564 1498 1.60 Abu 2 161 66 1233 43 74 232 0.23 Abu 3 111 45 1688 75 28 194 0.21 Subtotal 1303 524 57010 36 831 1775 0.70 HABU / NG8 2093 441 386912 61 817 3370 3.91 Total 3396 965 443922 39 2080 4711 1.42 Figure 20. Distribution of zebra in Abu (NG 26), and Habu, Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 26 NG 27a Fence NG 29 40
Table 23. Population estimates and statistics for birds Abu (NG 26), Oct10. Fish eagle Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 13 5 17 61 5 21 0.02 Abu 2 37 15 59 40 21 52 0.05 Abu 3 96 39 448 45 53 139 0.19 Total 146 59 524 31 101 191 0.08 Saddle-bill stork Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU / NG26 Abu 1 22 9 43 59 9 36 0.03 Abu 2 41 17 89 46 22 60 0.06 Abu 3 91 35 317 40 55 127 0.18 Total 154 61 449 27 112 196 0.08 Wattled crane Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) ABU/ NG26 Abu 1 25 10 132 92 10 48 0.04 Abu 2 34 14 86 53 15 38 0.05 Abu 3 89 35 549 40 41 136 0.17 Total 148 59 767 36 93 202 0.08 Bateleur eagle Stratum Estimate No. seen Variance Density % CI Lower CL Upper CL (km2) ABU / NG26 Abu 1 Abu 2 21 8 55 71 17 36 0.03 Abu 3 73 39 203 38 44 102 0.14 Total 94 47 258 66 62 126 0.05 41
Figure 21. Distribution of birds in Abu (NG 26), Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 27a NG 26 Fence NG 29 42
Table 24. Population estimates and statistics for cattle in Habu (NG 8), Oct10. Stratum Estimate No. seen Variance % CI Lower CL Upper CL Density (km 2 ) HABU / NG8 17339 3659 4228348 24 13120 21559 32.41 Figure 22. Distribution of cattle in Habu (NG 8), Oct10. NG 25 MOREMI GR Thaoge R NG 8 NG 26 NG 27a NG 8 Habu Fence NG 29 43