How the game census in the Kruger National Park, other than the buffalo and elephant census, is done, has changed significantly over the last 10 years. In essence the methodology has changed from a 'total' count to a sample survey technique, mainly due to staff and financial constraints as the strategic focus and management plan of the Park changed during the mid to late 90s.
Adverse climatic conditions also prevented the completion of 'total' surveys each year between 1994 and 1997, the last of these surveys being done in 1993. The first count of larger herbivores, excluding elephant, rhinoceros and hippopotamus, attempting total coverage of the whole area of Kruger from a fixed wing aircraft was done in 1977.
The method was refined and standardised to ensure repeatability. A team consisting of the pilot, a data recorder and four observers flew parallel strips 800m apart, the observers looking at an area 400m wide on either side of the aircraft, to achieve total coverage of the park. Flying was always limited to the cooler parts of the morning and in clear weather conditions.
But, the Park is just under two million hectares in size, and it meant the survey took three and a half months to complete each year, it was costly and time consuming, though considered effective in terms of creating understanding of the KNP ecosystem at the time.
In 1998 the Park adopted the 'distance sampling' technique to achieve population estimates for larger antelopes like impala, giraffe, kudu, wildebeest, waterbuck, rhino, zebra, steenbok and warthog. A 15 percent coverage of the Park was thought to be sufficient, but this was increased to 22 percent in 2001 and to 27 percent in 2004.
A fixed wing aircraft with four observers is still used, but the flight lines for the aeroplane are more widely spaced, and so cover less of the park. The observers also look at animals within 500m of the aeroplane.
Judith Kruger, data analyst, of the Kruger National Park explains: Equally spaced strips (transects) are flown from east to west across the entire park to obtain a 27 percent coverage. As the plane moves along the transect line the animals that are seen are recorded as well as their distance to the transect line. The sightings are made up to 500m on either side of the plane.
Distances recorded are used to develop a sighting curve based on the fact that one expects to see more animals close to the line than further away. The distances also correct for visibility bias or undercount bias which is a well known error in aerial counts and which was not taken into account in the historical 'total count'.
The sighting curve is then used to estimate densities of each species surveyed in the KNP. This technique allows one to compute certain statistical parameters that give us an indication of the accuracy and precision of the count as well as allowing us to include associated data in the analyses to increase the precision of the estimates.
This method did not involve repeated sampling for the purposes of calculating statistical data such as variances or confidence limits. The data from these counts were accepted at 'face value', although it was acknowledged that the 'total' count had not actually counted every single animal of a particular species in the park.
As species differ in their 'sightability' from the aeroplane, differing proportions of each species' population would have been counted. However, because a standardized technique was used it was accepted that these proportions remained relatively constant from year to year. It was also accepted that trends observed in census totals reflected actual population trends.