Code and data from: Survey-based inference of continental African elephant decline

Long-term quantification of temporal species trends is fundamental to the assignment of conservation status, which in turn is critical for planning and targeting management interventions. However, monitoring effort and methodologies can change over the assessment period, resulting in heterogeneous data that are difficult to interpret. Here, we develop a hierarchical, random effects Bayesian model to estimate site level trends in density of African elephants from geographically disparate survey data. The approach treats the density trend per site as a random effect and estimates a parametric distribution of these trends for each partitioning of the data. Data were available from 475 sites, in 37 countries, between 1964 and 2016 (a total of 1,325 surveys). We implemented the model separately and in combination for the African forest (Loxodonta cyclotis) and savannah (Loxodonta africana) elephant species, as well as by region. Inference from these distributions indicates a mean site-level decline for each species over the study period, with the average forest elephant decline estimated to be more than 90% compared to 70% for the savannah elephant. In combination, there has been a mean 77% decline across all sites; but in all models, substantial heterogeneity in trends was found, with stable to increasing trends more common in southern Africa. This work provides the most comprehensive assessment undertaken on the two African elephant species, illustrating the variability in their status across populations. 

This submission contains the code for the analysis of the survey data, which are included in the submission. Due to differences in resources for monitoring across sites and the development of new techniques over decades of data collection, surveys varied widely with respect to method, effort, and frequency. More specifically, the methodology and temporal range of data differed between sites; and at any one site, surveys may have used different methods, with different associated levels of observation error, and with different survey area sizes that may or may not have included the complete elephant population. We further lacked information on intrinsic demographic rates of growth or carrying capacity, which change across the continent due to environmental conditions. These limited and inconsistent data constrained our analytical approach in three important ways. First, we modeled elephant density rather than numbers since the survey area size was not constant over time for most survey sites. Second, we were able to fit only the simplest exponential population model: A logistic model of density-dependent growth did not converge. Third, we lacked overlapping, comparative data across sites that would allow us to calculate an overall measure of population change directly from estimated site-specific trends.