Data from: Designing effective oral rabies vaccination strategies for endangered carnivores: Lessons from bait uptake in Ethiopian wolves
Data files
Jun 01, 2026 version files 61.74 KB
-
Rabies_oral_vaccine_trial_data.csv
54.30 KB
-
README.md
7.44 KB
Abstract
Effective disease control is vital in mitigating devastating disease outbreaks among threatened wild animal populations living in proximity to domestic animal reservoir populations. With only ~500 individuals remaining in small, isolated populations in Ethiopian highlands, the most immediate threat to endangered Ethiopian wolves, Canis simensis, is rabies transmission from nearby domestic dogs, Canis familiaris. Since 2018, the Ethiopian Wolf Conservation Programme (EWCP) has conducted oral vaccinations of Ethiopian wolves to control rabies spread. Analysis of camera trap images from 580 bait stations across five Ethiopian wolf populations between 2018-2020, enabled species uptake to be confirmed for 416 baits. Ethiopian wolves were the main bait consumers (46%), with peak uptake occurring soon after deployment in the evening. Non-target uptake was almost exclusively diurnal, and predominantly by rodents, birds of prey and domestic dogs (41%). Generalised linear models showed that location and vegetation type influenced bait uptake by Ethiopian wolves and indicated that interference competition with domestic dogs can reduce Ethiopian wolf bait uptake near human settlements and where dogs are more active during the day. Our methods and framework offer valuable insights for designing effective oral vaccination strategies in wild carnivores globally, especially where domestic animal interference threatens wildlife health interventions.
Dataset DOI: 10.5061/dryad.mkkwh71g3
Description of the data and file structure
Files and variables
File: Rabies_oral_vaccine_trial_data.csv
Description:
NA is not available data
Variables
- Bait.no.: First bait deployed (1), or second bait deployed (2) at the station.
- Bait.site: Identity/name of the baiting station
- Cam.ID: Identity/name of the camera trap deployed
- Area: National Park or broad geographical area in which the oral vaccination took place
- Location: More local sub-population in which the oral vaccination took place
- Pack: Name of the pack whose territory the oral vaccine baits were deployed in
- V: Vegetation type, codes - 1 - Alpine grassland meadow; 2 - Alchemilla meadow; 3 - Helichrysum meadow; 4 - High altitudinal swamp; 5 - Erica moorland; 5a - Erica forest; 5b - Closed Erica moorland; 5c - Open Erica moorland; 5d - Early Erica regeneration; 5e - Degraded Erica; 6 - Open grassland; 7 - Kniphofia grassland; 8 - Artemesia grassland; 9 - Alchemilla haumanii; 10 - "Guassa" grassland; 11 - Charranfe heaths
- Date: Date of the oral bait deployment (dd/mm/yyyy)
- Bait.time: Time that the bait was deployed
- Bait.eaten.: Binary variables, was the bait eaten (Yes) or not (No)
- Check.time.bait.gone: Time period that the bait was checked in the field and was no longer present
- Check.time.gone.time: Time period that the bait was checked in the field and was no longer present
- Sachet.found.: Was the oral bait sachet found by the field team? Y = yes, N = No, YP = yes, partially
- Eating.code: Image tagging code related to an animal consuming the bait
| Behaviour | Evidence | Outcome (Positive (P)/Likely (L) uptake) |
|---|---|---|
| Eats (E) | Animal eats the bait | P |
| Eats and chews (EC) | Animal eats the bait and chews | P |
| Eats, chews and spits (ECS) | Animal eats bait, chews it and spits part out | P |
| Eats and spits (ES) | Animal eats bait and spits part out | P |
| Eats partially (EP) | Animal eats part of bait | P |
| Eats unknown item (EUNK) | Animal eats something that cannot be confirmed as the bait, e.g. rodent/attractant | L (if bait marked as gone at next check time in field data) |
| Bait displaced (D) | Animal moves bait out of camera view | L (if bait tagged as gone at next check time) |
| NA | No evidence of eating in camera images | L (if bait tagged as gone at next check time) |
-
Start.interaction: Timestamp of first image/video in the sequence in which an animal interacts with the oral bait
-
End.interaction: Timestamp of final image/video in the sequence in which an animal interacts with the oral bait
-
Date.of.interaction: Date of first image/video in the sequence in which an animal interacts with the oral bait
-
Day.eaten..0.1.2.: Day that the oral vaccine bait was eaten, day of deployment (0), one day after deployment (1), two days after deployment (2)
-
Species.in.interaction: Species interacting with the bait, codes: AB = Augur buzzard; AC = African civet; AGW = African golden wolf; AO = Abyssinian owl; BV = Bearded Vulture; CL = Caracal; CT = Cattle; DC = Domestic cat; DD = Domestic dog; DK = Donkey; EM = Egyptian mongoose; EW = Ethiopian wolf; GB = Gelada; HA = Hare; HB = Honey badger; HO = Horse; HY = Spotted hyaena; JB = Jackal buzzard; MU = Mule; OB = Olive baboon; SC = Serval cat; SF = Saker falcon; ST = Sheep & goats; TBR = Thick-billed raven; TE = Tawny eagle; WM = White-tailed mongoose
-
Bait.gone: Timestamp when the bait is gone (consumed, removed, out of view etc.)
-
Bait.interaction.length: How long the animal interacted with the bait (HH:MM:ss)
-
No..of.individuals.in.sequence.of.eating.: Number of individual animals present in the bait interaction sequence (for those involving any eating codes)
-
Uptake.status...L.: Outcome of whether the bait was consumed by an animal.
1. Positive uptake:
There was an eating code recorded within the bait interaction series, and the bait is marked as gone either in the camera data, or at the next check time in the field data. This includes the E, EC, ECS, EP, ES codes, but excludes the EUNK code (where it’s unclear what the animal is eating) or the D code (where the bait is displaced out of view from the camera); in these cases, the outcome would be tagged as likely.
2. Likely uptake:
The bait is marked as gone in a bait interaction series, but there’s no evidence of an eating code recorded in the series (unless EUNK or D tag).
-
Bait.unaccounted.for.: Whether the bait was unaccounted for in fieldchecks
-
Season: Season, wet or dry. Accounting for seasonal variation across the northern and southern sites, the wet season was defined as July-October and the dry season as November-February (Eshete et al., 2018; Sillero-Zubiri et al., 1995a).
-
Wolf.density: Kernel density estimates of space use (from Ethiopian wolf sightings) and focal pack numbers were used to calculate Ethiopian wolf population density for each study population in the relevant vaccination year.
-
Housing: Distance to the nearest human house (where the dogs live and return to at night), which was calculated for each bait station using the Distance to Nearest Hub function in QGIS (km)
-
DD.density: Kernel density estimates of domestic dog sightings from opportunistic sightings during Ethiopian wolf and general wildlife observations in Bale between 2018-2020 (n=496 sightings). The kernel was weighted by the number of domestic dogs seen per sighting.
-
Vheight: Vegetation height - categorised as Short (~0-10cm), Medium (~10-30cm), Tall (~30cm+).
-
VROD: Rodent density, as predicted by vegetation height (short vegetation = high rodent density, medium vegetation = low rodent density, tall vegetation = medium rodent density)
-
LOCATION: More local sub-population in which the oral vaccination took place - as used in analysis
-
Outcome: Oral bait outcome - consumed by an Ethiopian wolf (EW), consumed by another animal (Other), not consumed (NOT)
-
Village: Distance to the nearest human settlement (where the dogs live and return to at night), which was calculated for each bait station using the Distance to Nearest Hub function in QGIS (km)
Five Ethiopian wolf populations across three mountain ranges in Ethiopia were vaccinated between 2018-2020; South Galama in the Arsi Mountains (2 packs), Abuna Yosef in North Wollo (3 packs), and the Web Valley (Web – 8 packs), Sanetti Plateau (Sanetti – 6 packs) and Morabawa (8 packs) in the Bale Mountains (see Figure 1). Twenty bait stations were set up in each pack territory, each separated by 250-300m where possible. The bait stations were arranged in four parallel 1-1.2km transect lines (5 bait stations per transect), forming a grid structure.
Vaccination and camera trap survey
Sachets containing the SAG2 rabies vaccine were stitched inside goat meat baits (the preferred bait type for Ethiopian wolves; (Sillero-Zubiri et al., 2016)) and stored at -20°C prior to deployment. A single vaccine bait was placed out in the early evening of day 0 at each bait station and if removed, then a second bait was placed out the following evening (day 1). The vaccine needs to remain in a cold chain, so baits remaining on the morning of day 2 were discarded. Bait stations were regularly checked to document whether the bait was still present, and camera traps (RECONYX, HC600HYPERFIRE; Cuddeback, Attack BlackFlash; Browning, Prometheus) were set 5-7m in front of the bait to record the species removing the bait. Camera traps were motion-triggered. The vegetation type immediately surrounding the bait station was also recorded.
Camera trap image analysis
Camera trap images were analysed in order to identify the species removing the baits and the factors influencing bait uptake by target and non-target species. All camera trap images were tagged using digiKam software (digiKam team, 2021). Tags were extracted from the image metadata using ExifTool (Harvey, 2003) and R code, and subsequent data manipulation was conducted in R (v 4.1.1; (R Core Team, 2025)).
Images and videos were tagged at bait station level with the camera ID, location, year and pack territory. Date and time were automatically written to the image metadata and manually corrected if needed. Images and videos were tagged as a “bait interaction series” to help identify which species removed baits. A “bait interaction series” was defined as the sequence of images in which a carnivore or bird of prey appeared in the frame, or in which any animal interacted with the bait. Independent “bait interaction series” were defined as having an interval of 
15 minutes between images where an animal is present (based on a red fox camera-trapping study; Dorning & Harris, 2017). This interval was deemed appropriate because red foxes show ecological similarities to Ethiopian wolves; social canids displaying solitary foraging behaviour (Dorning & Harris, 2017).
In any given bait interaction series, the species and number of individuals present, along with the length and nature of the interaction, specifically: eating, chewing, spitting, sniffing or ignoring the bait were recorded. Animals were identified to species-level using morphological characteristics where possible, otherwise a higher-level classification was used, e.g. rodent, unknown carnivore or unknown bird of prey. The presence of multiple species in a single image/sequence of images was recorded as one “bait interaction series”, and the corresponding behaviour of each individual in relation to the bait was recorded. In order to confirm bait uptake by an individual, or to provide a time window for removal, the first image where the bait is clearly gone from view was also marked.