Recognising the fairly wide confidence limits on surveys, it is nevertheless surprising to find no inverse relationship between estimates in years when surveys were carried out both inside and outside the park It might be expected that in years when Etosha showed a relatively 'low' population a higher than usual estimate would be obtained outside the park.

Data on elephant mortality in Etosha dates back to 1971 (Lindeque 1988). Martin (2005) modelled the Etosha population in an attempt to resolve the relationship between numbers, mortality, immigration and emigration over the period 1971-2004. The performance of the population is consistent with one which has a central mortality of slightly more than 1% resulting in an intrinsic growth rate of 3.3%. Central mortality is defined here as the age specific mortality affecting all animals between the ages of 5-45 years. The schedule of overall mortality, immigration and emigration which follows from this is shown in Figure 13.

Results

The model results indicate that the population increased rapidly between 1971 and 1983 to its highest recorded level of 2,800 animals - mainly as a result of immigration although the mortality between 1971 and 1979 was lower than expected. This period coincided with a heavy hunting pressure outside the park and with a surplus in the cumulative rainfall as measured in Kamanjab, North Western Namibia (Table 4a, Figure 13). The cumulative surpluses and deficits in rainfall are the integral of all deviations from the mean. In 1980 mortality shifted to being higher than expected (mainly as a result of anthrax) but this negative effect was not sufficient to stop population increase.

After 1983 the population declined sharply due to a combination of factors of which the emigration of almost 1,000 animals in 1985 had the greatest effect. Two culling operations removed 570 animals from the population. 220 elephants were culled in 1983 and a further 350 in 1985. Mortality remained higher than expected up until 1990. In 1983, the cumulative rainfall switched to a deficit mode and the illegal hunting pressure outside the park was greatly reduced. These two factors may have resulted in the 1985 emigration referred to above.

Apart from a brief increase to a level of 2,000 animals in 1987 caused by the presumed immigration of some 600 animals, the population declined to 1,188 animals in 1995. The mortality during this period was not sufficient to have caused the decline and it must be attributed to a sequence of small emigrations between 1988 and 1995, none of which were statistically significant in isolation but which in concert reduced the population by some 800 animals.

After 1995 the population increased - again at a rate exceeding the intrinsic growth rate. Although this coincided with a period of lower than expected mortality, the increase must be attributed to immigration from 1996-1998. From 1999-2004 the population has fluctuated around 2,300 animals.

Conclusions

1. The findings of Lindeque (1988, p235-238: Conclusions) are corroborated from the model. Certain increases and declines cannot be explained through any mechanism other than immigration or emigration.
2. Mortality, even at the peaks of anthrax epidemics, has not been high enough to cause substantial declines or even to regulate the population.
3. The application of confidence intervals to the model data (Appendix 2) suggests that only a few of the postulated instances of migration are likely to be statistically significant when the survey results from one year to the next are compared. The corollary to this is that, whilst immigration and emigration do occur, the major migrations involving more than 50% of the population are episodic events which occur only once in every 10 years. Most migrations (if they occur) involve a small fraction of the population. According to the model about 80% of all annual immigrations and emigrations involve less than 20% of the population and more than 50% of these involve less than 10% of the population (Figure 11).
4. The cumulative surpluses and deficits in rainfall over the period concerned do not provide consistent evidence of a close relationship either with migration or mortality. The increase in the population up until 1983 coincides with a period of rainfall surplus. The sharp decline in the population after 1983 coincided with the transition from a cumulative rainfall surplus to a deficit. The higher than expected mortalities from 1983-1992 occurred during the period the rainfall was in deficit mode. However, thereafter the relationship falls apart. Despite being in a continued rainfall deficit regime after 1995, the population increased and mortalities were less than expected.
5. An unexpected outcome of the modelling process was the derivation of a unique value for the Finding Factor for elephant carcases (Appendix 2). According to the model 53% of all elephant deaths in Etosha are recorded.

Kamanjab Rainfall Data

Table 4a: Kamanjab Rainfall Data