Study area

We conducted GPT trials in the vicinity of the Ipassa Research Station in the Ivindo National Park, Gabon. This region of Gabon (Fig. 1) experiences bimodal rainfall with two dry (January–March and June–August) and two rainy (September–December and April–May) seasons, with a mean annual precipitation of approximately 1700 mm and mean annual temperature of 23.9°C (Koerner et al. 2017). Ivindo National Park is approximately 300 000 ha and is predominantly forested (Sassen and Wan 2006). We selected this study area as, at the time of the study, there were five GPS-collared forest elephants present within the park (two in the north and three in the south).

Focal elephant

We intended to incorporate multiple GPS-collared elephants in the study. However, only one elephant – a female named Amelia (age 30–40) – reliably ranged near the research station during the study period. Amelia was collared on 24 March 2017 by the Agence Nationales des Parcs Nationaux. Camera trap footage and dung size distributions (Turkalo 2013, Turkalo et al. 2018) demonstrated that Amelia was consistently accompanied by a seven or eight years old juvenile, Porkchop and a two–three-year-old juvenile, Rosebud. Both camera trap footage and the size distribution of dung-piles encountered suggested that the group remained together during the study period, therefore we assumed that the juveniles mirrored Amelia's movements.

Experimental seeds

Four types of experimental seeds were selected to mimic natural seeds of different sizes to measure elephant gut passage time. Two of the four experimental seeds were ‘passive’ – used to mark the point of defecation by the elephant. Passive seeds have been used previously to determine the gut passage time of a variety of captive vertebrate species where defecation time can be directly observed (Sadeghayobi et al. 2011, Steuer et al. 2011, Bunney et al. 2017), whereas here we infer the gut passage time of free-roaming, wild individuals from dung location and GPS collar data. Two of the seeds were ‘active’ – they recorded the actual gut passage times on the device through ambient temperature fluctuations.

The smallest ‘passive’ experimental seeds were colored plastic pellets (6 mm diameter Airsport Ammo; 0.11 g) – from here on referred to as ‘pellets’ (Fig. 2A). Different colored pellets were employed for each baiting trial to distinguish the origin of recovered pellets.

The second type of ‘passive’ seeds consisted of hollowed out Chrysophyllum lacourtianum seeds filled with aluminum foil and sealed with medical safe epoxy (Henkel Loctite Hysol M-31Cl Medical Device Epoxy) – from here on referred to as ‘tin seeds’ (Fig. 2B). Chrysophyllum lacourtianum is a large tree with large succulent globose fruits (mean mass = 320 g) that contain three to five seeds (mean mass = 2.8 g, mean length = 29 mm, mean width = 14 mm) and is consumed by elephants in the region (Rosin and Poulsen 2017). We harvested seeds from ripe fruits collected under wild fruiting trees. Insertion of the aluminum foil made it possible to detect the tin seeds in elephant dung using a metal detector (Garrett Ace 150 Metal Detector). Once hollowed out and filled with aluminum foil, the mass of the tin seeds was reduced by approximately 2.25 g on average relative to the unmodified seeds. To discriminate the origin of different tin seeds, a soldering iron was used to mark each seed with a unique pattern of dots corresponding to each baiting event.

Figure 2.

Photographs of the experimental seeds from above (top row) and the side (bottom row). Where: A=‘pellet’; B=‘tinned seed’; C=‘iButton’; D=‘VHF tag’.

The first ‘active’ experimental seeds consisted of iButton thermocouples (Thermochron DS1921H-F5, Maxim Integrated) (17.4 mm in diameter, 4.5 g) encapsulated in medical safe epoxy (Henkel Loctite Hysol M-31Cl Medical Device Epoxy) for protection – here on referred to as ‘iButtons’ (Fig. 2C). iButtons can be calibrated to record ambient temperature (with an accuracy of ±1°C) at defined time intervals (here we used 60 s) and have unique identification numbers so that each device can be individually identified. Therefore, after consumption by an elephant, the length of time in which the recorded temperature is consistent with the body temperature of an elephant (∼36°C/96.8°F) is roughly equivalent to the GPT. Previously, Kinahan et al. (2007) fed iButtons embedded in apples to captive African savanna elephants to estimate GPT. We performed a similar baiting trial with three captive savanna elephants at the North Carolina Zoo to test the effectiveness of our encapsulation methods (Supplementary material Appendix 1).

The second ‘active’ experimental seed was similar to the first with the addition of a VHF transmitter (R1170, Advanced Telemetry Systems) (34 mm in diameter, 4 g) to improve the probability of recovering iButtons from free-roaming, wild elephants – here on referred to as ‘VHF tag’ (Fig. 2D). The VHF tags emit a signal that can be detected with a hand-held receiver and antennae. The maximum detection distance when the VHF tags were embedded in elephant dung was 120 m in the open and 58 m in the forest.

Baiting

We tracked Amelia's movements in real-time between 11 June 2018 and 2 August 2018 using an smartphone application developed by save the elephants (STE) and Vulcan, Inc. To focus our baiting efforts, we identified places Amelia visited repeatedly from the GPS tracking data and selected six elephant baiting locations (Fig. 1). At each location, we cleared the ground of vegetation and attached two camera traps (Bushnell Natureview Camera Viewer Model 119740 or Bushnell Trophy camera HD Brown Model 119676) to nearby trees, one aimed at the bait on the ground to directly observe feeding events and the other aimed at approximately 1.5 m in height to aid with individual identification (as non-collared elephants occurred in the area).

When Amelia and her juveniles approached the baiting stations, we selected the three most likely sites and baited them with all four experimental seed types embedded within real fruit. The fruit pile consisted of wild C. lacourtianum fruits, a seasonally and locally abundant elephant-dispersed fruit, and market bought mangos (Mangifera spp.) and ripe plantains (Musa spp.), both of which are known to be consumed by crop-raiding elephants in the study region. Each C. lacourtianum fruit was hollowed out, embedded with up to six tin seeds, no more than one iButton or VHF tag, and stuffed with a mixture of C. lacoutianum and plantain fruit flesh, and colored pellets. Each plantain contained a single epoxied iButton or VHF tag and was also stuffed with the fruit/pellet mixture. Prepared fruits were washed with stream water to reduce the possibility of seed rejection through human handling. In total, each pile contained 250 g of pellets (∼2270 beads), between 19 and 46 tinned seeds and up to five iButtons and VHF tags. Additional fruit, such as mangos, undisturbed C. lacourtianum and plantains were added to each pile to make them more attractive to the elephants.

Once the sites were baited, we monitored the STE app periodically to see if Amelia approached the active bait locations. After a potential feeding event (the focal elephant coming within 50m of the bait pile), the feeding stations were checked for signs of fruit consumption. If the fruit piles were disturbed, we reviewed the camera trap footage to determine if Amelia or her juveniles had consumed fruit. If they did, we began an ‘elephant follow’.

Elephant follows

To recover dispersed seeds, we followed the focal elephant's GPS track and searched for dung piles. We conducted the follows a minimum of one hour behind Amelia to avoid altering her movement behavior. An experienced local Baka tracker followed the physical trail of the elephants through the forest. During our first follow, we tracked Amelia from the initial baiting station, and the first seeds were recovered 18 h after ingestion. On subsequent follows, we initiated tracking from Amelia's GPS location 10 h after fruit ingestion. We continued follows for 10 days after fruit ingestion.

For every elephant dung encountered, we measured the circumference of the three largest boli to identify the age of the individual. All boli were hand searched for experimental seeds, then spread into a thin layer and swept with a metal detector. For each dung pile with experimental seeds, we recorded the location and counted the number and type of seeds recovered.

Measuring GPT

In order to estimate gut passage time of passive experimental seeds, we developed a moving window algorithm in the R statistical environment. Briefly, for each dung encountered the elephant track of the focal elephant was subset to only contain the movement data from the time of feeding for the first dung encounter, or 6 h before the previous dung encounter for all subsequent dung, to the time the dung was discovered. We then determined the nearest available GPS point from the track subset to the located dung pile. The difference in time from the initial feeding event to the time at the nearest elephant location to the dung pile was considered to be the GPT (Supplementary material Appendix 2).

Passive experimental seeds: pellets

Overall, we recovered 842 pellets across the three follows (215, 534 and 93 in each follow respectively). The average number of pellets found per dung pile during follow one was seven, with a minimum of one and a maximum of 60. During follows three and four, the number of pellets were not counted to exhaustion, as the counting process would have compromised our ability to remain in close proximity to the focal elephant, therefore searches were limited to ∼2 min. The rate of recovery diminished with increasing GPT, plateauing at around 60 h (Fig. 3). Estimates of GPT based on pellet-containing dung piles and GPS location data resulted in a GPT distribution ranging from 16.6 h to 79.7 h. The mean GPT was 39.7 h (SD = 14.5 h) (Fig. 4).

Passive experimental seeds: tin seeds

Three out of 48 tin seeds were retrieved from dung piles, two on follow three and one on follow four. Two of the tin seeds were found in two dung piles that also contained plastic pellets, with GPT's of 29.2h and 53.8h. The third tin seed was found by itself in dung, and had a GPT of 113.7h. The mean GPT calculated using tin seeds was 65.6h.

Figure 3.

Cumulative number of pellets over estimated gut passage times, in hours, for three elephant follows.

Active experimental seeds

We did not recover any iButtons or VHF tags on the elephant follows (despite similar experimental seeds being recovered in trials with captive savannah elephants; Supplementary material Appendix 1). This is likely because Amelia spat out the encapsulated seeds shortly after consuming them. During the first baiting trial we recovered six iButtons that had been spat out on the ground at the baiting site, leaving only two that were potentially ingested. The unlocated seeds could also have been spat out after leaving the feeding station. During the third baiting trial, the elephants (Amelia and/or her juveniles) spat out five tin seeds, four iButtons and one VHF tag. We did not recover any rejected active seeds from the feeding site of the fourth trial, most likely due to the presence of tall grasses making them difficult to relocate. Our fifth and final baiting trial was unsuccessful, Amelia did not swallow any of the experimental seeds. She investigated the fruit pile and chewed some fruits, but threw most fruits away upon smelling them.

Figure 4.

The frequency distribution of gut passage time (GPT) in hours from dung piles with plastic pellets. See Supplementary material Appendix 3 for the frequency table.