Assesment of strip transect and Distance sampling on aerial surveys of elephants in Kruger National Park

“There is no ideal method for counting elephants” (Barnes et al. 1999:5). With the exception of those few areas where total ground counts have been accomplished by identifying every individual in the population (i.e. Amboseli and Addo), elephant numbers must be estimated. Aerial surveys are often used in savanna habitats, but ground counts may also be made on foot or from vehicles. Aerial total counts and sample counts are the two survey methods most commonly used to inventory elephants in eastern and southern Africa, and only a few areas use dung counts (Barnes et al. 1999). Total count surveys are commonly used in many Kenyan and South African reserves. Although these counts are considered “relatively accurate”, they are logistically expensive to conduct and rarely is there an independent assessment of their accuracy. Further, they are considered a total census, thus, there is no measure of their precision. Douglas-Hamilton (1996) provided a detailed description of this survey method. Aerial sample counts are also commonly used. Traditionally, the strip transect method pioneered by Norton-Griffiths (1978) has been used. Similar to total counts, there is rarely an independent assessment of the accuracy of this survey method; however, the population variance is typically calculated using Jolly’s Method II (Jolly 1969). Mbugua (1996) provided a detailed description of this survey method.

Although the strip transect method is widely used, this method has several assumptions that may frequently be violated, and thereby bias the population estimate. Most importantly, this method assumes that all objects (elephants) in the strip are observed (Buckland et al. 1993). Yet, seasoned elephant observers readily admit that some elephants, albeit few, are missed by observers on aerial surveys (I. Douglas-Hamilton, H. Jachmann, G.C. Craig, pers. comm.). This may be inconsequential if the goal of the monitoring program is only to provide an index to abundance. However, the estimate will be biased if the goal is to obtain an accurate estimate of the elephant population. DISTANCE sampling allows a relaxation of this strong assumption that all objects are detected by developing a detection function for the survey data (Buckland et al. 1993). Thus, a major goal of this project is to evaluate the reliability and performance of DISTANCE sampling relative to strip transect sampling for estimating elephant abundance using aerial surveys. It will also provide wildlife managers an important tool for developing accurate and efficient elephant survey programs, thereby strengthening their national elephant conservation efforts.
OBJECTIVES:

1. Conduct systematic aerial sample surveys of elephants in Kruger National Park to evaluate the accuracy and precision of DISTANCE sampling and strip transect methods in comparison to an aerial total count.

Traditionally, strip transects have been used to survey elephant numbers throughout southern and eastern Africa. However, Conservation International (CI) used DISTANCE sampling in their surveys in the Ngamiland District of northern Botswana in 1999/2000 (Burm and Griffin 2000) and in the Kafue National Park of Zambia in 2001 (Fairall and Kampamba 2001). In their Botswana surveys, Burm and Griffin (2000) reported that estimates of elephant abundance derived using DISTANCE sampling were 7 to 31% higher than estimates derived using strip transect estimators. Yet, the relative accuracy of these two estimators is unknown because an independent total count of elephants in their study area was not available.

Since 1967, Kruger National Park has conducted total count surveys using helicopters, providing accurate long-term data on KNP elephant numbers.
In 2001, park staff included elephant bulls (not family herds) in their fixed-wing census of other larger herbivores using DISTANCE sampling (I. Whyte, pers. comm.). With a 22% coverage of the park, they obtained an estimate of 1,343 bull elephants (1,023 < n < 1,764, α = 0.95). This compares to their total count of 1,313 from their helicopter census. Thus, DISTANCE sampling may hold promise for elephant surveys, but this survey method needs more experimentation for both bulls and breeding herds. A major concern is an under-sampling of juveniles because they are obscured by the adults. Potentially, a correction factor based upon recruitment rates could be used to compensate for this bias (I. Whyte, pers. comm.).

The purpose of this project is to assess the reliability and performance of DISTANCE and strip transect sampling methods for surveying elephant numbers. The accuracy and precision of these two sampling methods will be evaluated for the Kruger National Park elephant population for which an accurate total count using helicopters is available. This project will provide resource managers with information for determining which survey techniques may be most appropriate for their elephant conservation programs. It will also clarify which survey technique should be used for the large region-wide surveys planned by CI for northern Botswana during 2002/2003.

2. Evaluate a variety of factors that affect DISTANCE sampling counts, including: animal movement, observer, and vegetation type.

A wide variety of factors can affect the accuracy and precision of population estimates obtained from aerial surveys (Norton-Griffiths 1978). Thus, it is critically important to minimize any potential sources of bias. Additionally, if it occurs, it is important to estimate the magnitude of these sources of variance so that they can be accounted for in the abundance estimate. The flight transects will be stratified according to several vegetation types to determine if detection probability varies by vegetation. We will also use double observers, a video camera and the multiple covariate analysis feature in DISTANCE 4.0 to monitor for the occurrence of bias and to maximize precision of the abundance estimates. This assessment will help to identify the potential sources of bias in DISTANCE sampling surveys, and strategies for minimizing their effect on abundance estimates. It will also provide resource managers a tool for planning more accurate and precise aerial surveys for elephants.

3. Develop a technical manual detailing DISTANCE sampling protocols for elephant surveys, including survey design, equipment specifications, observer training, survey methods, and data analyses.

Data on elephant abundance and density are the cornerstone of all elephant conservation programs (Dublin and Taylor 1996). Whichever counting method is used, it is important that counts are reliable and repeatable. The manager also needs to know something about the quality of the data. Development of a manual on the reliability and performance of DISTANCE sampling and the protocols for designing, conducting and analyzing such surveys will help resource managers decide which sampling method to use in their elephant conservation programs.