Edited by: Timothée Vergne, Ecole Nationale Vétérinaire de Toulouse, France
Reviewed by: Richard Anthony Kock, Royal Veterinary College (RVC), United Kingdom; Gustavo Machado, North Carolina State University, United States
This article was submitted to Veterinary Epidemiology and Economics, a section of the journal Frontiers in Veterinary Science
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Bushpigs (
In many resource-constrained countries, livestock and wildlife may interact due to the weak biosecurity in subsistence farming systems, abundance of wild species populations, and overlap between farmland, forest, and protected areas. This situation may lead to the transmission of multi-host pathogens shared between livestock, wildlife, and potentially humans, affecting the livelihoods of communities (
African swine fever virus (ASFV) is a multihost pathogen that is able to infect several species of
Among social research methods, the questionnaire, aimed at recovering knowledge from stakeholders such as farmers, hunters, wildlife rangers, or livestock traders, is considered a practical, fast, and cost effective approach to gather information regarding interactions between wild and domestic animals possibly in a large area and over long periods (
Direct observations of animals or records of presence indicators (e.g., tracks and fecal droppings) have also been used. They are efficient and usually demand limited resources. Nevertheless, they depend on suitable field conditions, trained personnel and are often time-consuming. They rarely allow a continuous monitoring of the animals but may result in a coarse estimation of the spatial and temporal pattern of the interactions (
Camera-trap (CT) is a non-invasive tool for estimating interactions at previously selected spots, allowing simultaneous monitoring of several species and providing behavioral information (
Unlike observations and CT, telemetry does not require preliminary knowledge on the locations of wildlife-livestock interactions. By following the animal movements, this method provides spatial and temporal data on the overlap between the monitored individuals (wild and domestic) and livestock facilities at a fine-scale. This approach yields a substantial amount of data needed to determine and characterize the interface (
Therefore, the choice of one or a combination of several methods usually depends on the specific objective of the study. For instance, investigations on wildlife-livestock interactions and disease transmission should take into account the pathogen of interest, the infected hosts, as well as, their ability to transmit the pathogen through different routes. In those cases, the selection of the method may be constrained by the accessibility to the animals, and the inherent requisite resources. Subsequently several studies have previously used different methods in parallel or consecutively to gather complementary or preliminary data at the interface (
In this study, we investigated an
The study was carried out in Nwoya District (total population: 138,500; area: 4,736 km2) located in northwestern Uganda. The study site comprises the northern boundary of an unfenced protected area, the Murchison Falls National Park (NP), and the adjacent rural communities. Specifically, the study included 23 villages of the southern part of the district, covering about half (2,600 km2) of the entire district (Figure
Study site and results yielded by the three methods for each crop field included in the sample for the rainy and the dry season. The camera symbol locates the fields where bushpig visits were detected by CT. Footprints and circles spot the results drawn by the questionnaires and the track observation as described in the legend.
The study area is covered by various land items ranging from built up areas, bush, grassland, subsistence farmland and woodland (National Forest Authority). The climate is tropical with a rainy season that runs from April through November; and a dry season from December to March. The area was strategically selected for this study due to recurrent occurrence of ASF outbreaks among a growing free-range domestic pig population (
In order to determine the influence of certain drivers such as land use and distance from the crop fields to the park boundary on the frequency of visits by the bushpigs, we designed a systematic sample scheme based on 54 regular points, each point being spaced 5 km from each other, covering different types of land use and within a distance of 20 km from the Murchison Falls National Park limit. This distance was chosen under the assumption that bushpigs are more abundant inside the park or at the boundary than further away from the park, and that their movement can stretch to a maximum distance of 15 km (
The spatial selection of the farms in the study was based on the sample scheme. Farms were then selected by convenience to meet different criteria such as agreement of the farmer, type of grown crops, safety for the CT (we avoided crop fields that were not regularly visited by the farmer or exposed to bush fires) and logistic constraints with a target sample size of 30 farms. Twenty-eight farms were finally included in the study. Two farms which were originally part of the sampling scheme declined participation and could not be replaced within the time frame of the study. The total number of crop fields owned by the 28 farmers was 145. The distribution of the different crops owned by the farms is shown in Table
Type and number of crops grown by the 28 interviewed farmers.
Cassava | 20 |
Sesame | 19 |
Maize | 16 |
Groundnut | 14 |
Sorghum | 14 |
Rice | 14 |
Soya bean | 12 |
Peas | 11 |
Bean | 10 |
Sweet potato | 7 |
Millet | 5 |
Sugar cane | 2 |
Cabbage | 1 |
The farm was the sampling unit for the questionnaire survey while the crop field was the sampling unit for both the CT and the tracks observations surveys. Within each farm, crop fields to be monitored for the CT and track surveys were chosen to have a balanced sample among the different available crops and the landscape variables. When possible, we monitored the same field during the dry and rainy season. However, some of the crops monitored during the rainy season were harvested at the dry season. In this latter period, we selected, when possible, crops that were not yet harvested within the same farm. As a consequence, some of the crop fields differed between the rainy and the dry season. The distribution of the types of crops that were surveyed according to the seasons is shown in Table
Distribution of the crops monitored according to the season and their stage (P, planting; Mi, middle; Ma, mature; H, harvested).
Cassava | 13 (3 Mi + 10 Ma) | 9 (3 Mi + 6 Ma) |
Maize | 3 (1 Ma + 2 H) | 4 (1 Mi + 3 Ma) |
Groundnut | 1 (H) | 3 (Ma) |
Sorghum | 2 (Ma) | 2 (1 Mi + 1 H) |
Sweet potato | 3 (1 P + 1 Ma + 1H) | 1 (Ma) |
Soya bean | 1 (H) | 2 (Mi) |
Sesame | 2 (H) | 1 (P) |
Rice | 1 (H) | 2 (1 Mi + 1 Ma) |
Bean | 1 (H) | 1 (Mi) |
Millet | 0 | 2 (1 Mi + 1 Ma) |
Peas | 1 (H) | 0 |
Sugarcane | 0 | 1 (Mi) |
Cabbage | 1 (Mi) | 0 |
Permission to carry out the study was granted by the Ugandan National Council for Science and Technology under the reference number A497. A written consent from the District veterinary officer was obtained prior to the start of any activity in the area. At the time of the interviews, participants were informed that the study was voluntary, confidential, and that they had the choice of ending their participation at any time. An informed consent was given by all participants prior to the implementation of the study.
The questionnaire, designed for individual interviews, consisted of 115 questions. The questions were designed to collect data related to crop raiding by wild and domestic animals. It was reviewed by a local and international team of epidemiologists and social scientists and uploaded in the KoBo toolbox (KTB) online platform (
The questionnaire survey was implemented from November 2016 to January 2017 and farmers were inquired about data corresponding to the dry and rainy seasons pertaining to the year 2016. After having described their farms in terms of crops grown, they were asked about the frequency of visits from wildlife and domestic animals and the season when the visits occurred. We also asked them to rank the species in descending order from the one causing most damage to the one causing the least. Complementary data on the geographical characteristics of the farms were gathered using QGis 2.10. Pisa (Table
Description of the variables used in the camera-trap and tracks surveys analyses.
Response variable: frequency of bushpig visits | Ordinal variable, 5 classes |
Number of bushpig visits detected per 24 h and per session | Presence of bushpig tracks per session (0/1) | |
Explanatory variables | Crop ( |
ground | ||
Distance from the crop field to the nearest forest | ground | |||
Distance from the crop field to the park boundary | NFA/GIS | |||
Distance from the crop field to the nearest river | NFA/GIS | |||
Land use ( |
NFA | |||
Season ( |
Date |
We used 6 infra-red motion triggered cameras (Trophy Cam, Bushnell Outdoor Products, USA) which detect movement within a 15 meters range, have a trigger speed of 1 s and display 32 infrared night vision LEDs. The cameras were set to record pictures and 10 s video footages each time a movement was detected within the distance range.
A total of 41 crop fields were selected (mean: 1.46 ± 0.58 crop fields per farm). For each selected crop field, one camera was placed in a site where tracks of wildlife were previously observed by the track observer (see next paragraph). When such observations were absent, the camera was placed where wildlife foragings were more likely to occur, such as the edge of the field bordered by forest or bush, at the opposite end of human settlements. Cameras were tied on a tree or a stalk either at the average bushpig height (~30–50 cm above the ground) or higher (150–200 cm above the ground) with a downward pointing inclination, depending on the surrounding environment. To prevent non-specific triggering of the cameras due to movement of the surrounding vegetation, the grass and branches that fell in the field of view of the CT were removed.
The cameras were set to work continuously during day and night. Date and time were displayed for each photo and video captured. We deployed the cameras from June 2016 to April 2017 in order to cover the rainy (from June to the end of November 2016 and April 2017) and the dry (from December 2016 to the end of March 2017) seasons. We defined a session as a 10 days continuous period of monitoring on the same place with the same camera and position. After each session, the cameras were rotated, so that each crop field was monitored for one session within the rainy season and one session during the rainy season. The location of the camera was recorded using a handheld GPS unit (Garmin GPS Map 60Cx).
All videos and photos were read for species identification and count of the number of individuals. From the photos and the video footages, we recorded the behavior of the animals, such as foraging or just passing through the crop field. We defined independent visits as (1) consecutive photographs or footages of individuals of different species, (2) consecutive photographs or footages of individuals of the same species more than 30 min apart, or (3) non-consecutive photographs or footages of the different or same species (
Each crop field where the CT was set was investigated twice for wild animal tracks by the same trained field assistant to ensure standardization of the data: once when the camera was set and subsequently, when it was removed at the end of the session. These two observations were recorded during the dry and the rainy seasons as for the CT survey. Each time, the whole area of the field (1.2 acre in average) was scanned to physically look for any animals track. The observer was not the same person as the one watching the video footages from the CT, allowing independence between the two methods. For each observed set of tracks, the species was identified, based on the footprints, droppings, and type of crop damage. When the identification was not possible at the species level, the group (such as “antelope”) was recorded.
In the questionnaire, the different ordinal qualitative modalities of the frequency of bushpig visit (never/at least once in a cropping season/at least once in a month/at least once in a week/at least once in a day) were transformed into ordinal numerical values (0/2/12/52/365), corresponding to the estimated minimum number of bushpig visits reported per year in the crop field. This was the response variable. The number of visits by bushpigs yielded by CT (quantitative variable) and the presence of tracks (dichotomic variable, 1: present/0: absent) were the response variables for the two other methods, respectively. Given that tracks could remain visible for several days, depending on the seasonal conditions, and that the session was planned to last 10 days, we assigned the value of 1 when the tracks were present either at the first or second observation or both.
The explanatory variables were the season, the type of crop, the distances from the crop field to the nearest forest (we took the distance between the location of the CT and the nearest forest taken by a handheld GPS as described above), to the nearest park boundary and to the nearest river. These two latter distances were calculated between the locations of the camera and these features obtained from administrative shapefiles layers plotted on QGIS. Other variables included land use (bush, grassland, woodland, farmland) obtained from the National Forest Authority (NFA) land cover shapefiles layers (2008) and plotted on QGIS (Table
From the questionnaire, a descriptive analysis conveying the responses dealing with bushpig visits was carried out. From the CT survey data, we calculated the number of bushpig visits per night per session and the mean and standard error of the visit frequency. From the track observations, we calculated the number of sessions in which the bushpig tracks were observed.
Secondly, we analyzed how the three response variables varied among the type of crop, the distance of the crop field to the forest, park boundary and river; and also between the dry and rainy seasons. We did this by using generalized linear mixed models (GLMM) to take into account the likely dependence within one farm. For the questionnaire and CT survey data sets, we used a Poisson link as usually used for count data having Poisson distribution (
Model selection was performed following the procedure described by Zuur et al. (
In order to compare the results from the questionnaire, for which the sampling unit was the farm, with the CT and track survey for which the sampling unit was the crop field, we extracted questionnaire data corresponding to the crop fields monitored by the two other methods. We also transformed the variables corresponding to the visit frequencies given by CT and the questionnaires into dichotomic responses (presence/absence of bushpigs) for each crop field and each season.
We evaluated the degree of agreement between the results obtained by the different methods by computing the Kendall coefficient of concordance for each season, using the irr R package (
We also performed a GLMM to investigate the ability of the three methods to detect the bushpig presence in relation to the drivers highlighted by the models selected for each method separately. We did this by including the method (questionnaire/CT survey/track survey) as a qualitative variable in the fixed effects. The response variable was the presence or absence of bushpigs for one crop field with a given method. In order to test whether methods performed differently depending on the situation investigated, we included the interactions between the method and the different drivers found to have an effect on the frequency of bushpig visits. We used a binomial link and the farm was set as a random effect. The selection procedure was the same as already described above.
These analyses were performed using R 3.4.2 with the same packages as described in the previous paragraph.
Finally, we compared the three methods regarding the time and the cost (in terms of manpower and material) that were required to achieve the study. We also compared the characteristics of the data obtained from their collection to their processing leading to the variable of interest.
Three quarters of the farmers (21/28) reported visits and crop damage from bushpigs. Among them, only three reported seeing bushpigs in their field during daytime. All the other reports were based on the evidence of bushpig tracks. When asked how they differentiated bushpig tracks from domestic pig ones; six respondents answered that bushpig footprints were bigger than the ones of domestic pigs; five replied that domestic pigs were absent of the area; four that the tracks were in too close vicinity of the bush to be attributable to the domestic pigs; four gave other answers and two were not able to make the difference. No direct interaction was mentioned by any of the respondent.
A total of 43 crop fields, out of the 145 owned by the farmers, were concerned by bushpig visits. Farmers provided different estimates of the visit frequencies depending on the crops they grew. Thirty seven percent of the crop fields (16/43) were visited by bushpigs at least once a day, 58% (25/43) at least once a week, and 5% (2/43) at least once a month.
The selected model for frequency of bushpig visits measured by questionnaires predicted that bushpig visits occurred more often in cassava and groundnut, than in the other surveyed crops. According to this model, the closer the field was to the NP boundary, the more frequent bushpigs forayed into the field (Table
Models selected to explain the frequency of bushpig visits and presence of bushpig tracks in crop fields yielded by questionnaires, CT or tracks observation.
Questionnaire | Frequency of | |||||
bushpig visits | Cassava | Ref | ||||
Groundnut | 0.79 [0.51–1.23] | 0.312 | ||||
Sweet potato | 0.43 [0.23–0.78] | 0.008 | ||||
Maize, sorghum, millet | 0.49 [0.33–0.72] | <0.001 | 0.453 | 0.616 | ||
Other | 0.04 [0.02–0.07] | <0.001 | ||||
0.50 [0.27–0.87] | 0.015 | |||||
Camera-traps | Frequency of bushpig visits | Data did not allow to select a model | ||||
Tracks | Presence/absence of bushpig tracks | 7.6 [1.72–2540] | 0.122 | |||
0.381 | 0.765 | |||||
Dry | Ref | |||||
Rainy | 11.3 [1.69–2286] | 0.068 |
Among wildlife, the bushpig was the most frequent reported species to raid the crops. This was followed by the African elephant (
Fifty-eight sessions were planned but the data from one of the sessions could not be retrieved due theft of the SD card in the camera. The remaining 57 sessions were distributed as followed: 28 during the dry season and 29 during the rainy season and recorded a total of 692 “camera-days.”
Sessions lasted 12.1 ± 5.2 days on average. The variation in session duration was due to loss of battery power or logistic constraints.
One visit was excluded from the analysis because we could not ascertain the identification of the species.
A total of 16 visits from bushpigs were visualized on the pictures and/or video footages, yielding an average frequency of 0.014 visits per day ± 0.05. The 16 visits were distributed among 5 crop fields belonging to 5 different farms, with a range of 1–7 visits per crop field per session. Six visits occurred during the rainy season, (3 in groundnut fields and 3 in cassava fields) while the 10 others occurred during the dry season, in cassava fields (Figure
No model could be selected from the CT results due to a limitation on the amount of data (only 5 fields) (Table
Many other wild species were visualized in the pictures and/or video footages, making a total of 75 visits. The most frequent group were the antelopes including dik dik (
Among domestic animals, goats were the most frequent with 32 visits. This was followed by domestic pigs (19 visits). Eighteen of the visits by domestic pigs occurred during the dry season, and nine were observed in cassava crop fields, eight in a maize crop field and two in a soya bean crop field. The visit occurring in the rainy season was detected in a cassava field, where bushpigs also came. The time interval between the two occurrences was 3 days.
According to the protocol, the number of sessions was the same as for the CT survey with a total of 58 sessions. The mean interval between the two observations of the same crop field was 13.8 ± 4.5 days. Bushpig tracks were detected in 21 of the 58 sessions (36.2%), distributed among 18 different crop fields (out of the 41 monitored) and 15 farms (out of 28). The selected model showed an effect of the distance from the crop field to the nearest river, tracks being more detected in farms located further to rivers, although this effect was not significant. The rainy season tended to be more favorable to the bushpig visits (Table
CT detected presence of bushpigs in 9% of the monitored crop fields, whereas the questionnaire and track survey detected bushpig presence in 39 and 37% of the fields, respectively.
The Kendall coefficients assessing the concordance among the three methods were 0.62 (
To test the hypothesis that methods performed differently according to the situation investigated, we included the interactions between the method and the variables found to have an effect in the GLMM selected for each method, i.e., distance field-park, distance field-river, and season and crop. The crop*method interaction was not included as it did not allow to correctly estimate the parameters.
No interaction was retained in the selected model. As for the questionnaire model, bushpigs more often visited cassava, groundnut, and sweet potato fields than the other crops and intruded more frequently in the fields located closer to the park. Similar to the tracks model, bushpig presence was associated with longer distance between the field and the river; and bushpig presence occurred mostly during the rainy season. Questionnaire and track observations reported bushpigs much more often than the CT method, whereas no significant difference was shown by the model between questionnaire and track observations (Table
Model selected to explain the bushpig presence in crop fields, using combined data from three observation methods.
Presence/absence of bushpigs | |||||
Cassava | Ref | ||||
Groundnut | 1.77 [0.31–11.42] | 0.524 | |||
Sweet potato | 1.39 [0.24–7.48] | 0.703 | |||
Maize, sorghum, millet | 0.10 [0.02–0.35] | < 0.001 | |||
Other crops | 0.04 [0.01–0.17] | < 0.001 | |||
0.56 [0.32–0.92] | 0.029 | 0.585 | 0.585 | ||
2.14 [1.37–3.49] | 0.001 | ||||
Dry | Ref | ||||
Rainy | 5.40 [2.12–15.12] | < 0.001 | |||
CT | Ref | ||||
Questionnaire | 14.67 [4.25–63.31] | < 0.001 | |||
Tracks | 13.09 [3.80–56.14] | < 0.001 |
The Table
Time and cost spent in euros to implement questionnaire, CT and the track surveys in the 2600 km2 study site during one dry and one rainy season, achieve the data collection and analysis.
Required time | ||||
Estimated cost (for the whole study) | ||||
Characteristics of the data | Sampling unit | 1 farm with at least 1 crop field | 1 crop field | 1 crop field |
Temporal sampling | Exhaustive | 10 days/season | 2 observations/season | |
Nature of the data |
Replies to questions |
Pictures and video footages |
Track observations Observer skills |
|
Data processing from the data collection to the building of the database | Automatic through Kobo software. Extraction and recoding of the data of interest | 1. SD cards collection, pictures downloading, visualization and filtering |
1. Data sheets filling |
|
Variable used to quantify the visit of BP | Number of reported BP visits in the farm's crop fields per year Categorical variable | Number of detected BP visit per day in the crop field |
Absence or presence of BP track in the crop field per session |
From these results, we draw the strengths and weaknesses of each method according to our study. They are summarized in Table
Strengths and weaknesses of the three methods used to study bushpig visits in crop fields.
Strength | - Cost |
- Can provide quantitative and behavioral data | - Cost |
Weakness | - Lack of specificity |
- Cost and time-consuming |
- Need skills to identify the tracks |
There are very few scientific publications about bushpigs compared to other species of wild pigs. This is probably due to their nocturnal habits and elusive nature. In this study, we were interested in providing information on the most efficient methods to monitor buhspig incursions into crop fields, as an indirect measure of the potential for direct or indirect interaction with domestic pigs, and thus the potential for disease transmission. We used three different methods (questionnaire, CT, and track survey) to assess visits by bushpigs to crop fields belonging to farms surrounding Murchison Falls NP in North western Uganda. Cassava, groundnut, and sweet potato were the crops associated with the highest frequencies of bushpig visits. Proximity to the NP boundary also had a high correlation with bushpig visits to farms. Bushpigs were more often detected during the rainy season and in fields further away from a river. Consistency between methods was better in the rainy than in the dry season. Questionnaire and track observations reported much more bushpig intrusions into crop fields than CT.
The number of farmers reporting bushpig visits in their crops (75%) are consistent with what was found in a nearby district (Masindi district) by Hill (
The influence of the distance between the farm and the NP boundary suggests that bushpigs might be more abundant at the edge of the park than further away. No data are available regarding the bushpig density outside and inside the park, which could support this result. In Kenya, however, Okoth et al. (
The effect of the distance to the river is less clear. Bushpigs are known to require access to water and habitat with sufficient moisture to support dense vegetation throughout the year (
In accordance with our results, Kukielka et al. (
According to the Kendall coefficient, concordance among the three methods was fairly good during the rainy season and intermediate during the dry season. The fair agreement in the rainy season could be explained by a more accurate assessment of bushpig foray by the farmers in this season due to regular visits in their field to cultivate the crops. Moist soil also makes observation of the tracks more accurate because footprints remain for a longer time and allow a more reliable identification of the species. However, it is worth noting that the selection of the model including the three methods did not result into any effect of the interaction between the method and the season, reflecting that the method efficacy was not influenced by this factor. It is noteworthy that combining the results of the three methods allowed us to obtain a more powerful dataset with better estimation of effects.
Although CT was the most specific method allowing to identify the species with more certainty, visualize the number of individuals and their behavior and provide a quantitative estimation of the bushpig visits, it was also the least sensitive in detecting their presence. The number of visits by bushpigs drawn by the CT might have been underestimated for several reasons. First, we used only one camera per crop field and could have missed bushpig incursions occurring outside the field of view of the camera. Secondly, animals passing rapidly in front of the motion sensor without staying in the field of view may trigger the camera but could not be “captured” even if the trigger speed was set to minimum (1 s), leading to “false negative” results (
Data recorded from the CTs and the track observations were collected concurrently but independently (i.e., the tracks observer was not aware of the data collected by the camera-trap on the same field and in the same session). We had lower photographic rates than track rates, just as Silveira et al. (
In the questionnaire study, bushpig forays were more frequently reported by farmers than the results yielded by CT and track surveys. These observations may not be fully reliable since they are based on indirect observations (tracks) and not direct sightings. Moreover, the farmers' ability to identify tracks from bushpigs was heterogeneous and mismatching between bushpigs, warthogs and domestic pig tracks is likely to have occurred, although we cannot say to which extent. Another factor to take into account is that people may be less tolerant toward wildlife than toward livestock regarding crop damages because they have very limited control over wildlife activities (
Interestingly, the questionnaire and track surveys highlighted different risk factors (type of crop and distance to the park boundary for questionnaire, season and distance to the river for tracks survey). This result reflects the usefulness of the two methods to identify spatial and temporal hotspots for bushpig presence in farmland and the potential for bushpig-domestic pig interaction in our study area. We did not find any interaction between the method and any of the risk factors tested. This means that there was no interference between any of the three implemented methods and the spatial and temporal factors we tested. However, regarding the very low sensitivity and the high cost of camera-trapping in comparison to the questionnaire and track surveys, we would recommend to use one of the two latter methods to study this type of interface.
Our results showed that bushpigs visit crop fields and that some of these areas may be more at risk of intrusion depending on their location and type of crops grown. Eventhough pig farming is common among the rural community of this area, the domestic pig population is not evenly distributed and is directly related to the distribution of the pigs owners. Mapping the farms keeping domestic pigs would be of particular interest to see how they overlap with bushpigs' preferred field locations, although pig keeping is not a permanent activity and this map would need to be regularly updated. The CT survey detected 19 visits from domestic pigs, most of them (18) occurring during the dry season and nine of them in cassava crop fields. Pigs are usually tethered or housed during the rainy season to prevent them from feeding in the growing crops. This practice may reduce the potential for interactions with bushpigs since we showed that they forayed into crops mostly during the rainy season. Pigs which are not tethered during this season are much more at risk to interact with bushpigs, particularly in cassava fields, which are attractive for the two species. The only observation of consecutive visits by the two species in the same field within a short time interval (3 days) was collected by CT in a cassava field during the rainy season. Our data also showed that domestic pigs might visit crop fields at night, increasing the potential for direct contact with bushpigs. The use of GPS or VHF technology would be of great value to investigate these shared habitats between both species. This tool allows to monitor animal movements at regular and possibly very close intervals, providing a clear picture of the animal's home range and use of the habitat (
In our study, it was noteworthy that warthogs were not frequently reported or detected in crop fields by any of the implemented methods. This result suggests that crop fields are not a common space of interaction between warthogs and bushpigs or domestic pigs.
The choice of a method to study wildlife-livestock interface may be constrained by the spatial scale, the accessibility of the animals, the availability of funds, human resources, and time. Here, questionnaires and track observations were most relevant to describe the frequency of bushpig visits and their determinants. The results of this study confirm the interest of using crop field as study area of interaction and pathogen sharing between these two species. Such interactions may lead to the transmission of a number of shared pathogens other than ASFV, such as helminthosis or bovine tuberculosis (
AP and PO designed the study and collected the data in the field with the help of AO who was the local facilitator and the contribution of CM, KS, and FJ. AP, EG-F, and EE performed the statistical analysis. AP wrote the manuscript. AP, KS, CM, and FJ conceptualized the thrust and focus of the manuscript. AP, EG-F, CM, KS, EE, and FJ participated in drafting the manuscript or revising it critically for content.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We acknowledge all the farmers who took part in the study and made that work possible. We also thank Erika Chenais from SVA for her valuable comments on the questionnaire, David Chavernac from CIRAD for his help in using KoBo toolbox Christophe Ferrier from ONCFS for his help with the figure and Esther Kukielka from UC Davis for her help with the data. We are grateful to Raymond who drove us and took part into the field work, Philip, Brenda, and Kevin, the team in Lutuk, who help with the logistics.