1. Over the past 50 years, aerial counts have been widely used in African wildlife management; however, the accuracy of the resulting estimates has rarely been questioned. The reliability of aerial counts of large African herbivores was examined by comparing the results of a series of systematic aerial sample counts with those from a series of line transect foot counts conducted in the Lupande Game Management Area in Zambia.
2. For most large herbivore species, the estimates from the aerial counts were considerably lower than those from the ground counts. The data pointed to undercounting as a major problem of aerial surveys. During the aerial counts, significant numbers of animals were missed on the transects, first due to the low probability of spotting single animals, small groups of animals and less conspicuous ones (sighting probability bias), and secondly because part of the population was concealed by obstructions and therefore not visible to the observers (visibility bias).
3. The main factors that influence visibility of large herbivores from the air are the animals’ reactions to an over-flying aircraft, dispersion, body size and colour. Animals that move in response to an aircraft are more likely to be seen than static ones; dark-coloured animals are easier to spot than light-coloured ones against a light background; large herds are easier to detect than small ones; and large animals are more easily seen than small ones. Body size is important while trying to spot grazers and mixed feeders from the air, while colour is important for spotting browsers. This is mainly due to the differences in habitat use, with browsers being confined to the thicker habitat.
4. To minimize undercounting bias, both conventional aerial counts and aerial line transect counts should be restricted to large conspicuous grazers and mixed feeders in medium to large group sizes, such as elephant Loxodonta africana africana, buffalo Sincerus caffer, zebra Equus burchell, wildebeest Connochaetes and lechwe Kobus leche. Operational factors, such as height, speed and strip width, should be kept within reasonable limits for conventional aerial counts. Only one species should be counted at a time, always applying the double-count technique. For aerial line transect counts, the use of a helicopter is a prerequisite to obtaining accurate estimates. Other important considerations are observer experience and flight duration.