Premium
Understanding the drivers of mortality in African savannah elephants
Author(s) -
Schlossberg S.,
Gobush K. S.,
Chase M. J.,
Elkan P. W.,
Grossmann F.,
Kohi E. M.
Publication year - 2020
Publication title -
ecological applications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.864
H-Index - 213
eISSN - 1939-5582
pISSN - 1051-0761
DOI - 10.1002/eap.2131
Subject(s) - african elephant , geography , poaching , population , ecology , population density , ecosystem , population growth , demography , biology , wildlife , sociology
Abstract Populations of African savannah elephants ( Loxodonta africana ) have been declining due to poaching, human–elephant conflict, and habitat loss. Understanding the causes of these declines could aid in stabilizing elephant populations. We used data from the Great Elephant Census, a 19‐country aerial survey of savannah elephants conducted in 2014 and 2015, to examine effects of a suite of variables on elephant mortality. Independent variables included spatially explicit measures of natural processes and human presence as well as country‐level socioeconomic measures. Our dependent variable was the carcass ratio, the ratio of dead elephants to live plus dead elephants, which is an index of recent elephant mortality. Carcass ratios are inversely proportional to population growth rates of elephants over the 4 yr prior to a survey. At the scale of survey strata ( n = 275, median area = 1,222 km 2 ), we found strong negative associations for carcass ratios with vegetation greenness at the time of the survey, overseas development aid to the country, and distance to the nearest international border. At the scale of ecosystems ( n = 42, median area = 12,085 km 2 ), carcass ratios increased with drought frequency and decreased with human density and overseas development aid to the country. Both stratum‐ and ecosystem‐scale models explained well under one‐half of the variance in carcass ratios. The differences in results between scales suggest that the drivers of mortality may be scale‐specific and that the corresponding solutions may vary by scale as well.