Selecting Contexts

Text to be written

Defining & Coding Behaviors and Constellations

The original behavior databases and Poole & Granli (2011) and Poole (2011) described some 218 behaviors and 47 calls that occurred in specific contexts. Each behavior was placed in a Narrow Context within eight overall General Contexts and described and referenced each behavior. To keep the database from becoming too complex each behavior was put in one narrow context that represented the "best fit" except behaviors that were also observed in play, in which case the word Play was added as a pre-fix, (e.g. Play-Mount, Play-Kneel-Down, Play-Bush-Bash, Play-Throw-Debris). Thus, Throw-Debris and Play-Throw-Debris were essentially counted as two separate behaviors.

In developing The Elephant Ethogram we recognized the short comings inherent in the choices we had made in the original database. We dropped the Narrow Context and instead describe 22 Contexts which we feel best cover the behavior of elephants. Having described a particular Behavior or Constellation we included it under each of the 22 Contexts in which it has been observed to occur. Thus, each behavior is described once. The only exception is behavior toward dead elephants, which are preceded by the term Body, as in Body-Mount, Body-Lifting etc. In the written description of each behavior we attempt to describe the various contexts in which the behavior occurs and, when it varies across contexts, the age and sex categories that we have observed, or expect, to engage in the behavior within each context.

We have tried to stick to behavior names that are descriptive (i.e. J-Trunk, Tusk-Clank, Reject-Suck, Sashay, Kick-Back) rather than use terms that make assumptions about what these behaviors signal - although there are a number that do fall into this category (e.g. High-Fiving; End-Zone-Dance, Casual-Stance).

We have given all behaviors that have the same appearance the same name. Thus, for example, what used to be called Periscope-Sniff (Poole & Granli, 2011) and Distant-Frontal-Attitude (Kühme, 1972) are now combined and referred to as Periscope-Trunk. Our aim has been to ensure that there is no overlap between the appearance of Behaviors, but in some cases (e.g. Circus-Pose, Periscope-Trunk, Trunk-Sweep, and Reaching-High) the differences may be subtle. [is this what I mean]. So while we have lumped Behaviors that had previously been given different names when they occurred in different contexts, we have split Behaviors with subtle but consistent differences that we are able to detect.

Each Behavior or Constellation is allocated a page on which we include a written description and video examples of the behavior. We have attempted to include video examples of all the contexts in which the Behavior occurs. We include at least one photograph of each Behavior, and, where relevant, audio examples.

For each Behavior or Constellation entered into the database we have coded in the following searchable information: 1) the context(s) in which the behavior has been observed to occur (selected from the following 22 Contexts); 2) the age and sex categories of elephants that have been observed to engage in the behavior (adult female, estrous female, adult male, musth male, juvenile adolescent female, juvenile adolescent male, calf, infant); 3) the part of the body actively used to engage in the behavior (entire body, ears, eyes, forefeet, genitals, head, hindfeet, mouth, tail, temporal glands, trunk, tusks); 4) the likely mode(s) of communication (acoustic-vocal, acoustic non-vocal, chemical, seismic, tactile, visual). Some of the behaviors are signals (e.g. Let’s-Go-Rumble; Ear-Folding), others apparent gestures (e.g. Trunk-to-Breast), some displays (e.g. Musth-Walk; Exaggerated-Foot-Splashing) and others are just things that elephants do (Drinking, Fell-Tree, Dig-Minerals) but each are associated with sounds, smells, movement or vibrations that may carry information to other elephants. For each Behavior or Constellation we have indicated what mode(s) of communication we believe other elephants might use to gain information about the sender. In some cases (e.g. acoustic-vocal for vocalizations) this is obvious, whereas in other cases (e.g. various foraging techniques) this may be conjecture.

Mining of footage


In a collaboration with copyright owners Off the Fence (The Netherlands), Gorongosa Media Project (US/Mozambique) and Bob Poole Films, hundreds of hours of raw footage of elephant behavior, originally shot for documentaries, were granted to ElephantVoices for use in science and education. Together with our own footage, photographs and audio files, these media put us in a unique position to document the behavioral repertoire and cultural diversity of elephants, as well as how environmental factors (such as habitat and human interference) may affect its expression.

The raw video data included 8TB of third party footage from Maasai Mara, Kenya, 5 TB of third party footage from Gorongosa National Park, Mozambique, as well as our own footage which included  1 TB from Maasai Mara, xx from Gorongosa and xx from Amboseli. Since we were carrying out research in each of these populations which involved individual identification, we were familiar with the elephants in the vast majority of the video examples. Known elephants are mentioned by name or ID code, otherwise they are noted by age, sex and characteristics.


Photographic examples of behavior were selected from our collections, which include xx GB from Amboseli, xx GB from Gorongosa and xx GB from Maasai Mara. Photographs from each population are stored in sub-folders by date, time and number of elephants, matching the group sightings data that we collect in the field. Images are viewed in Adobe Bridge and key-worded with individual elephant IDs and, where relevant, with Behavior and Constellation names.


Take from Poole 2011

Mining of footage


Data mining is defined as a process used to extract usable data from a larger set of any raw data. In this case we mined raw footage, photographs and audio recordings for examples of any recognized behaviors or behavioral patterns. In the process we discovered new behaviors and patterns of expression. We searched through and “mined” xx TB of footage from three elephant populations. The footage was imported into a number of Adobe Premiere Pro projects including Mara, Gorongosa 2011, Gorongosa 2012, Gorongosa 2013, Gorongosa 2014 and Gorongosa 2015, and our own material into Amboseli 16 Amboseli 19 Amboseli 20, Mara, and Gorongosa 17, Gorongosa 19. Within each project the footage was organized by date, by camera, and placed in a folder called Footage. From each day, camera or in the case of our own footage, each individual group, timelines or “Stringouts” were created. Each of these timelines could be linked back to group sightings and individual ID data and field notes that were collected on the elephants.

In a separate "Behavior" folder timelines were prepared for each previously known and described behavior (see Poole & Granli 2011; Poole 2011). To this original set we added new behaviors as we observed them in the footage. We scrolled through each timeline in the Stringouts Folder frame by frame, looking for examples of visible Behaviors or Constellations. Using Markers we carefully described the individuals and behaviors involved, noting their ages and sexes and individual identity, where known. Each time we detected a particular behavior, we copied the selection onto the relevant Behavior timeline. Since many of the Behaviors filmed involved a complex series of interactions between individual elephants, we typically copied the same selection, and its Marker description, onto several different timelines. The process of going through the material took many months over several years to complete.

Editing behavior clips


In 2018 Thomas Stafford began to go through the Behavior timelines that Joyce had built up, to begin the process of editing them for use on the database. He created a new Edit folder into which he, likewise, added timelines for each of the behaviors that Joyce had populated with exemplars. He then copied and pasted the best (cinemetography-wise) examples into these Edit timelines. He made each example into a potential educational clip including the Behavior-Name, circles and slo-mo to focus the viewers attention on the specific behavior, ElephantVoices logo, and footage credits. For Behavioral Constellations he added the individual Behaviors as they occurred in the clip.

Populating the The Elephant Ethogram


As the behavior clips were being prepared for export, we entered all information (Behavior or Constellation Name, Context, Population, Filename) regarding each of the over 2,000 clips into a FileMaker Database. We wrote a detailed caption for each clip, describing the context of the behavior, the individuals engaged in it and what specifically took place. The caption was entered into the FileMaker Database and also included with the Filename in a marker over the selected part of the timeline. Each clip was then exported in H264 1080x xxx .mp4. format and stored on our server.

Prepared clips, together with their captions and the context in which the behavior took place were uploaded to ElephantVoices Vimeo-account and, simultaneously linked from there to The Elephant Ethogram on The date of upload and the Vimeo link were entered on the same line in the FileMaker Database.

The Elephant Ethogram database

The concept and structural design of The Elephant Ethogram was developed by Joyce Poole and Petter Granli of ElephantVoices. Programming and technical maintenance is handled by Derrick Joel, Nairobi, Kenya in collaboration with Petter Granli. The Elephant Ethogram is coded in open source software PhP and MySQL in Joomla! CMS, and is hosted on Cisco servers. Links to video host Vimeo and audio host SoundCloud are handled through respective API.