Savannah elephants live in plains, thick bush and deep forests.
Elephants walk faster than people run. A fast-moving herd, even with youngsters, can walk fifty miles in one day.
Pilots often spot living and dead elephants during low flyovers. These two elephants have been dead for one day.
Most African governments have limited money for conducting counts. The Tanzanian government relies on its game scouts to chart the GPS coordinates of the live and dead elephants they see. Some countries also use dung counts or hire pilots to do low-altitude fly-overs.
The UN counts dead elephants through an arm called CITES. CITES runs a program called MIKE, and its members collect information on the number of elephants killed in certain regions. In 2013 Julian Blanc from MIKE said, “We are monitoring 30% to 40% of the elephant population through a peer reviewed process that gives us the best available global estimates on the illegal killing of elephants” (“Transnational”).
From early-2014 to mid-2016, an international team funded by Microsoft co-founder Paul Allen produced the Great Elephant Census. Its goal was to count all the African savannah elephants (not the forest elephants). At $7 million, the aerial survey covered the savannah elephant region in eighteen countries. So far, they have counted 352,000 elephants, which they say includes 95% of all African savannah elehpants ("Great").
No matter how one counts elephants, it is hard to get an accurate number because elephants live not only in the plains but also in the thick bush where it is hard to see them, and they walk many miles in one day, making it hard to know if some have been counted twice or not at all. Most importantly, the final numbers may be misleading. During any large-scale fly-over, “only about 6% of the actual area is surveyed. . . The population estimate and crucially, the estimated margin of error, are then calculated statistically from this sample count.” Thus the results are best presented in ranges, such as “between 11,077 and 20,595 elephants” (“South Africa”). This means that the results of any one count are imperfect. On the other hand, they do provide a benchmark for making comparisons from place to place or from year to year.