ANALYSIS

RESULTS

Africa

Using data collected by patrol staff at MIKE sites, the MIKE Programme estimates poaching trends based on the ‘Proportion of Illegally Killed Elephants’ (or PIKE). PIKE is calculated as the number of illegally killed elephants found divided by the total number of elephant carcasses encountered (see below for more details).

The charts below show levels in the Proportion of Illegally Killed Elephants in Africa and in Asia since 2003.

Africa Sub-regions

Asia

Estimating PIKE

The following video provides an overview of the MIKE data analysis.

However, this elephant mortality data provided by MIKE sites requires further work and analysis to be useful in developing a continent-wide picture of the number of and trends in illegally killed elephants. The challenge to generating reliable and robust elephant mortality information is that the conditions are different in every site, and, importantly, the percentage of the actual total number of dead elephants discovered and recorded will vary between sites.

For example, if a site is carrying out intensive patrolling, staff are likely to discover all or most of the poached elephants, suggesting a high poaching level. While if a site has a very low patrolling effort, they will only discover a small fraction of the poached elephants, suggesting a low poaching level. In reality, however, the reverse could be the case. Other factors, such as habitat type (e.g. forested verses savanna) also impact the likelihood that an elephant carcass will be discovered.

To address this issue, after more than a decade of practical experience in generating this information, the MIKE Programme, advised by its Technical Advisory Group (TAG), has developed a method for analysing site-level elephant mortality data, known as the Proportion of Illegally Killed Elephants, or PIKE. PIKE compares the number of illegal killed elephants to the total number of elephant carcasses found (from all causes of death, e.g. natural, legal and illegal).

In simple terms, the calculation of PIKE is the number of illegally killed elephant carcasses found at a site divided by the total number of all elephant carcasses found at the site over the same period (see below for more details on the actual methods used). As a ratio of the illegal elephant deaths to the total elephant deaths at a site during the year, PIKE values can range from 0.0 (i.e. all the carcasses found are from natural causes) to 1.0 (i.e. all carcasses found were illegally killed). A PIKE value of 0.5, for example, would mean that half the carcasses found were identified as illegally killed.

As the factors impacting the likelihood of discovering an elephant carcass (patrol effort, habitat type etc) are likely to impact equally on both illegal and natural deaths the resulting PIKE values should, in theory, be largely independent of these factors. As such, PIKE as an index of poaching levels, has been used in the MIKE analysis in an attempt to account for differences in patrol effort and other factors impacting detection rates between MIKE sites and within sites over time.

Biases in PIKE

There are a number of potential biases associated with PIKE that can result in either an over estimation or and under estimation of what the actual value truly is.

Over-estimating

Many experts agree that illegally killed elephants are more likely to be reported than those that die natural deaths. This could occur due to a number of reasons, for example:

• The detection of illegally killed elephants is often guided by intelligence or information from a wide range of sources (tourists, visitors, aerial surveys etc.). Natural deaths that occur in more remote or difficult to reach areas are more likely to be missed.
• Wildlife patrol staff typically focus their efforts in areas where poaching has historically been high. This makes sense from a law enforcement perspective but does mean that elephants that die from natural causes in other areas may not be recorded.
• In dense forest environments visibility is restricted limiting patrol staff ability to find elephant carcasses; and those that are found are likely to be in more accessible areas or on paths, which makes them more likely to have come into contact with poachers and to have been killed illegally.

Any of these factors would lead to an over-estimation of PIKE. This may be particularly the case in forest environments and where areas where a relatively low proportion of the total number of carcasses are actually reported.

Under-estimating

Conversely, there are also potential biases that could lead to an under-estimate of PIKE values. For example:

• Changing environmental conditions that increase the levels of natural mortality can result in a reduced PIKE while the actual number of illegally killed elephants could remain constant or even increase (for example, severe drought conditions have led to spikes in natural mortality in some areas in particular years).
• There may be perceptions among site staff that finding an illegally killed elephant could be viewed as poor performance. This could incentivise reporting carcasses which are not unambiguously poached as natural or unknown, or not to report an illegally killed elephant at all.
• Although only a fraction of the total number or illegal and natural elephant deaths are recorded, it is likely that the vast majority of elephants killed by management (e.g. to address human-elephant conflict) will be reported. This would in turn reduce the relative proportion of illegal deaths, therefore reducing PIKE.

The MIKE Programme, with support of its Technical Advisory Group, is continuing to work on possible ways to minimise these potential biases.

Elephant Mortality Data

Access to MIKE data is governed by the provisions of CITES Resolution Conf. 10.10 (Rev. CoP18). Summaries and aggregates of data provided to MIKE are provided in the links below.

• ​Numbers of carcasses found for each site and year
• Continental PIKE trend data for Africa
• Subregional PIKE trend data for Asia

​Anyone wishing to obtain more detailed elephant carcass data should get in touch with the CITES management authority of the elephant range State(s) concerned.

Analytical Code

The actual PIKE trend analysis is modelled using a least-squares means (LSMEANS) approach. The MIKE analytical method, and more specifically the programming code and description of its use for the PIKE trend analysis reported in document CoP18 Doc. 69.2 and Addendum, can be accessed in this GitHub repository.