Behavioural patterns of free roaming wild boar in a spatiotemporal context

Study area

The study area “Süsing” was a 17.8 km² state forest located in the Luneburg Heath in Germany. The region is characterised by large-area coniferous forest (Pinus sylvestris, Picea abies, Larix decidua & L. kaempferi, Pseudotsuga menziesii) and small-area oak (Quercus robur) and beech (Fagus sylvatica) forests (Keuling et al., 2013). The mean annual temperature is 8 °C and the average annual rainfall is approximately 700 mm (Keuling et al., 2013). Besides a high number of wild boar (about 8 animals/km² during the study period, calculated according to Rowcliffe via Random Encounter Model (REM) Rowcliffe et al., 2008), there are also high numbers of roe (Capreolus capreolus) and red deer (Cervus elaphus), as well as a few wolves (Canis lupus). Verbal permission was provided by Helmut Beuke (head of operations) at the Forestry Office of Oerrel to drive on to the closed forest roads and install the camera traps on their forest property.

Data collection

Direct observations (compared to radio telemetry) are required to record the behaviour of animals and consequently also get information on activity and habitat choices (Cagnacci et al., 2010). One cost-efficient method for the observation of free roaming wild boar is the use of camera traps. The advantage of camera traps is that they are non-invasive (Rovero, Tobler & Sanderson, 2010; Rowcliffe et al., 2011), and as a consequence, ideal to study nocturnal and crepuscular animals which avoid humans (Rovero, Tobler & Sanderson, 2010). The technique is applicable to the study of wild boar given they rarely react to camera traps (Amelin, 2014). Using ESRI^® ArcGis 10.1, 50 random points (Rowcliffe et al., 2008; Rovero et al., 2013) separated by a minimum distance of 100 m (Passon et al., 2012; Hofmeester, Rowcliffe & Jansen, 2017) were determined and afterwards explored with GPS (Rovero, Tobler & Sanderson, 2010; Rowcliffe et al., 2011). Additionally ten places with a high probability of wild boar occurrence (e.g., wallows, fresh rooting places, salt lick) were selected to reveal all behavioural elements necessitated for the ethogram. Cameras were placed at all 60 places and had an effective detection distance of 8–20 m. Animal’s behaviours were able to be defined in distances up to 20–30 m in front of the cameras.

The set-up of the Bushnell^® TROPHY CAM™ and Bushnell^® TROPHY CAM HD™ camera traps took place on 03.03.2014. The 50 cameras used for statistically evaluable behaviour observations were hung as near as possible to the random points, at trees in 90 cm height, orientated parallel to the ground (Rowcliffe et al., 2011; Meek, Ballard & Fleming, 2012) to capture some open space on the video clips, and if possible, a deer crossing which comes to or goes away from the camera (Bengsen et al., 2011; Rowcliffe et al., 2011). The additional ten camera traps, which are statistically irrelevant for the behaviour frequency, were hung at different heights (most of the time higher than 90 cm and with an angle <90° to the ground) depending on the area to capture a large field of view and thus a lot of behaviours. To not disturb the natural behaviour of the animals, no bait or lure were used at the random points (Rowcliffe et al., 2011; Meek, Ballard & Fleming, 2012). Each camera had a passive infrared sensor (PIR) and recorded, day and night, a 20 s video clip without sound when they were triggered. 1 s after the ending of the latest video the camera traps could be triggered again (Rovero, Tobler & Sanderson, 2010; Rowcliffe et al., 2011). The video clips were stored on SD cards, which were changed biweekly. Function of cameras and battery levels were checked during change of SD cards. After about three months, on 04.06.2014, the camera traps were retrieved.

The date and time for each clip was recorded (the time is presented in segments as full hours with the following full hour, e.g., 00:00 o’clock = 00:00:00 - 00:59:59 o’clock).

The habitat was described at each of the camera locations. First, every place was assigned to one of the six types: track, forest aisle, pond (incl. wallows), field edge (simultaneously edge of the forest), salt lick or wooded. After that, within a radius of 10 m the tree and shrub layer were described with main species (no trees/shrubs, broad-leaved, mixed or coniferous forest) and cover (0%, 0-50% or 50–100%). The herb layer was also divided as described. Here the main species were no herbs, common bracken (P.a. = Pteridium aquilinum), European blueberry (V.m. = Vaccinium myrtillus), bracken and blueberry (P.a.&V.m.) or herbs (e.g., Rubus sectio Rubus, Urtica dioica, Poaceae, a few Cyperaceae and Polypodiopsida). In addition, the cover of deadwood (0%, 0–25% or 25–50%) was registered.

In this study the sampling method “behaviour sampling” and the recording method “time sampling, one-zero sampling” (Altmann, 1974; Geissmann, 2002) were used. That means, during a sampling interval (video length of 20 s) all visible boar were observed as one group and it was noted for every behavioural element if it occurred in the video clip or not. An ethogram was created following literature review (e.g., Gundlach, 1968; Saebel, 2007; Briedermann, 2009) and own observations, at which exclusively the own observations are shown in Table 1.

Data analysis

Wild boar could be identified on 1,227 of ca. 8,500 video clips as well as at 57 of 60 places. From 1,169 video clips, a behavioural context could be analysed (645 of 673 video clips at the random points, 524 of 554 video clips at the other ten places), but only the video clips at the random points were statistically analysed, because the other ten did not fulfil the statistical requirements (not randomly, hung at different heights).

To compare the occurrence per day of the seven different behavioural categories at the random points, two analyses were done: First, to calculate the percentage of each behavioural category, the number of observations per behavioural element (BE) and random point (RP) at one day was calculated as the function: ${\displaystyle N_{\mathrm{obs},\mathrm{d}=1}\left(BE,RP\right)=\frac{N_{\mathrm{obs}}\left(BE,RP\right)}{d_{RP}}}$

where N_obs(BE,RP) is the total number of observations per behavioural element and random point and d_RP is the number of trial days per random point. Then the mean number of observations per behavioural element at one day could be calculated by: ${\displaystyle \bar{x}\left(BE\right)=\frac{\sum \left(N_{obs,d=1}\left(BE,RP\right)\right)}{50}}$ where 50 is the number of random points. The percentage for each behavioural category (BC) in% (with ∑(P(BC)) = 1) was then calculated by: ${\displaystyle P\left(BC\right)=\frac{\sum _{\left(BC\right)}\left(\bar{x}\left(BE\right)\right)\ast 100}{\sum \left(\bar{x}\left(BE\right)\right)}}$ where ∑_(BC)($\bar{x}$(BE)) describes the sum of the mean numbers of observations per behavioural element over all behavioural elements which belong to one behavioural category, and ∑($\bar{x}$(BE)) describes the sum of the mean numbers of observations per behavioural element over all behavioural elements.

Second, to compare the occurrence of the behavioural categories, the number of observations per behavioural category and random point was calculated as the function: ${\displaystyle N_{obs}\left(BC,RP\right)=\sum _{\left(BC,RP\right)}\left(N_{\mathrm{obs},\mathrm{d}=1}\left(BE,RP\right)\right)}$ where ∑_(BC,RP)(N_obs,d=1(BE,RP)) describes the sum of the numbers of observations per behavioural element and random point at one day over all behavioural elements, which belong to one behavioural category. Afterwards pairwise comparisons of means (over all random habitats, N = 300) were conducted with R software version 3.1.1 (R Core Team, 2014), using the packages “nlme” (Pinheiro & Bates, 2014) and “multcomp” (Hothorn, Bretz & Westfall, 2014). Therefore, the linear mixed model (LMM) (Dormann & Kühn, 2009) combined with the post hoc analysis least squares means (LSMEAN) (SAS Institute Inc., 2011) with Tukey adjustment (NIST/SEMATECH, 2013) was performed.

For the analyses of the behaviour in a spatiotemporal context, similar behavioural elements were grouped as listed: locomotion; sniffing and winding; defecating and urinating; vigilance behaviour; rooting and pawing; salt ingestion; sucking attempt and suckling; chewing and feeding (attempt); drinking; wallowing, nibbling and stretching; shaking; rubbing; scratching (one’s bottom) and rolling; social interactions; sexual behaviour (see Table 1). For each grouping, the number of video clips per time of day was summed over all 60 camera locations with Microsoft^® Excel 2007 to determine the activity maxima in general. Significant habitat preferences per behaviour were derived from a generalised additive model (GAM) dependent on the time of day and habitat type. Using the data from the random camera locations, it was calculated with R software version 3.1.1 (R Core Team, 2014), using the “mgcv” package (Wood, 2014), for each behaviour with greater than 20 observations. For the same data, tests for spatial dependence of residuals were conducted on a sample of 1,000 observations. We calculated Moran’s I for the first lag with R software version 4.0.2 (R Core Team, 2020), using the “ncf” package (Bjornstad & Cai, 2020). We did not find significant spatial autocorrelation. Since it is not possible to monitor the whole study area completely and consequently every possible habitat type, we can just draw conclusions out of the results given by random placed camera traps.

Behaviour in a spatiotemporal context

The maxima of locomotion and vigilance behaviour were observed with foraging behaviour (Fig. 2). Wild boar have to travel long distances while foraging and often have to cross open and unsecure spaces (Meynhardt, 1982; Cahill, Llimona & Gràcia, 2003), hence vigilance behaviour to avoid predators is important. Meanwhile the observed wild boar mostly used forest aisles or stayed in broad-leaved forests with a herb layer of 50–100%. In other studies it was found that wild boar preferred broad-leaved forest for foraging (Berger, 2006; Bertolotto, 2010). Wild boar move fast and take the shortest path when crossing an open unsecure space (Meynhardt, 1982). Manmade forest aisles that are rarely used by humans are probably used by wild boar (Allwin et al., 2016) to allow fast movement through forest areas. Thus, the hypothesis that foraging and related behaviour occur in broad-leaved forest is confirmed.

Social interactions and the only observation of sexual behaviour occurred during the maximum of comfort behaviour. This may be because comfort behaviour (Saebel, 2007) and social interactions could be observed mostly at the ponds (containing wallows) where the animals feel safe (Keuling & Stier, 2009). In addition, all ponds were located in coniferous forest, and wild boar prefer pine trees for rubbing (Mayer, 2009). Furthermore, these three behavioural categories could be observed many times at the saltlick. In general, however, nearly all of the seven behavioural categories could be observed at the ponds. Consequently, the hypothesis that comfort and related behaviour occur in coniferous forest and at places where the wild boar feel safe cannot be rejected.

If we compare the alternation of foraging and comfort behaviour during the night with the results of another study (Gundlach, 1968), the observations of the other study lack the first period of foraging behaviour after awakening and they also refer to diurnal wild boar. Our data support the results of other studies (Saebel, 2007; Keuling & Stier, 2009) which found that wild boar mostly attend to foraging behaviour in the first half of the night while a higher occurrence of comfort behaviour during the second half of the night is not obvious. It rather gives the impression that foraging is always followed by comfort behaviour. Consequently, comfort behaviour occurs later in the night than foraging behaviour.

Functions of the behavioural elements

The behaviour of an animal essentially contributes to its survival and reproductive success (Naguib, 2006; Kappeler, 2009). If we generalise the ecological model for the locomotion of wild boar (Morelle et al., 2014), it appears that the behaviour of wild boar is a result of the interaction of intrinsic (energy gain, escape from predators and/or conspecifics, reproductive success) and extrinsic (habitat, climate, presence of predators) factors - and thus, it is the struggle of wild boar with its biotope (Naguib, 2006). Our study supports this hypothesis. Further, we distinguished between basic animal behaviour serving the survival of the individuals and the sounder, and comforting behaviour aimed at the well-being of the individuals.

Our data supports, that wild boar use different behavioural elements for reaching different food resources. For example, rooting and pawing serve for the exposing of food sources in the ground (GÖT & BAT, 2003). Wild boar can distinguish between food places of different quality and relocate them which saves energy and time (Held et al., 2005). Moreover, sows suckle their offspring and therefore invest in the breeding and survival of their offspring (Vetter et al., 2016).

Our results show, that the functions of different behavioural elements are closely related. Wild boar, for example, have a very developed sense of smell (Graves, 1984; Mayer, 2009). The olfactory behaviour serves for foraging and avoidance of predators (sniffing and winding) as well as for intraspecific communication by defecating and urinating. Also rubbing, nibbling as well as nose-to-nose contact and nose-to-body contact serve for intraspecific communication.

Vigilance behaviour (pausing) seemed to be a reaction to the camera traps. Our results show that wild boar, compared to other animals, hardly react to camera traps, but when they react, they do it by eye-contact or pausing (Amelin, 2014). Wild boar are reclusive animals (Gundlach, 1968; Beuerle, 1975; Altmann, 1989). Vigilance behaviour is used by wild boar to avoid predation (e.g., by humans or wolves), for example when a sow guards a glade before other sows and young animals follow her. When pausing or laying down, the movement is abruptly stopped which otherwise would produce a noise, which predators could hear. Moreover, young boar are very camouflaged while laying down due to their striped pattern (Briedermann, 2009). The animals also use this moment to scan their environment multisensory (Quenette & Desportes, 1992). If the boar do not find the source of the noise or sense disturbing them, it could be that they react with flight.

The behavioural category comfort behaviour mostly serves for two functions, personal hygiene behaviour and resting behaviour. Looking at the personal hygiene behaviour, wild boar use wallowing for thermoregulation because they are not able to sweat and a mud layer also keeps stinging insects away (Meynhardt, 1982; GÖT & BAT, 2003; Briedermann, 2009). According to another study, wild boar immobilise stinging insects with help of the mud and afterwards remove them by rubbing and similar behaviour (Mayer, 2009). Rubbing is also caused by hair change in spring-time (Briedermann, 1971). Thus, comfort behaviour serves for the well-being of the animals in general (GÖT & BAT, 2003). In contrast to the results of previous studies, where stretching was always observed after resting behaviour (Briedermann, 2009), in our study stretching also could be seen three times mostly after rubbing and before shaking.

The social interactions of wild boar have different functions. Nose-to-nose contact and nose-to-body contact serve as intraspecific communication (see above). This is important for the mother-infant-relationship (Gundlach, 1968; Meynhardt, 1982), and for sexual behaviour (e.g., courtship of boars, boar fights), which serves for reproduction. It is said that each behaviour is noticed by group members and has social consequences (Stolba & Wood-Gush, 1989), allowing them to learn from each other (Schneider, 1980; Briedermann, 2009; Morelle et al., 2014). Young boar train from an early age on fighting and copulation in a playful manner (Gundlach, 1968; Meynhardt, 1982; GÖT & BAT, 2003), which they use later during the mating season for boar fights and mating. Wild boar also compete for food, however, they have a stable food hierarchy (Beuerle, 1975; GÖT & BAT, 2003; Saebel, 2007) to avoid unnecessary competition and to save energy.

Resting behaviour like sleeping was not observed in this study. Wild boar rest at their daytime resting sites (Gundlach, 1968; Meynhardt, 1982) which were never placed in front of any of the 60 camera traps. As wild boar prefer dense vegetation for their resting places (Allwin et al., 2016) it is statistically unlikely to catch such places randomly, since the camera traps need some open space to work correctly (cf. data collection). Again, we also do not know any resting place of wild boar in our study area, consequently it was not possible to place one of the 10 additional camera traps at their daytime resting sites. To analyse this behaviour in following studies we suggest permanently placing recording video systems at preferred resting sites which should be determined before with help of telemetry (Lampe, 2004; Sándor et al., 2014). Since our results stem only from videos in forest habitats, a lack of observations from open areas may explain lacks of activity maxima in the hour of 18:00 o’clock and between 01:00 and 02:59 o’clock, because at that time wild boar were probably on greens, fields or at baiting stations (in surroundings of private hunting grounds) for foraging. Another possibility is that the animals had an activity break between 01:00 and 02:59 o’clock in which time resting behaviour could have been observable. It has already been suggested that free roaming wild boar have a rest period in the second half of the night (Briedermann, 1971), diurnal wild boar around midday respectively (Allwin et al., 2016).

The expansion of humans results in wild boar’s habitat reduction. Due to the lack of natural predators in many places and increasing food supply, the wild boar population numbers are constantly increasing (Massei et al., 2014) and consequently it comes to their invasion into urban areas (Kotulski & König, 2008; Toger et al., 2018; Conejero et al., 2019). This leads to many conflicts between wild boar and humans. As wild boar are very adaptable, one method alone is not sufficient to reduce the animal’s number. In addition to the procedures already known (cf. West, Cooper & Armstrong, 2009; Tack, 2018), the results of this work show further possibilities, such as hunting the animals during their activity times at night with night vision devices at known social locations, and avoiding additional foraging resources (e.g., access to food waste) during their foraging activity times. In general, knowledge of habitat preferences and behavioural needs are useful for habitat management. Keeping wild boar in their “comfort habitats” could reduce human wild boar conflicts such as crop and rooting damages, if enough preferred habitats are available. Additionally, the public needs to be better informed about the effects of increasing wild boar population numbers, as there is, for example, a growing negative public opinion towards hunting (Tack, 2018). Consequently, wild boar behaviours drive the human perception of the wildlife-human conflict and thus determine the way of implementing wildlife population management measures. But we are also responsible for the living conditions of wild boar in captivity such as zoos and domestic pigs in factory farming. Enclosure design and activities can significantly improve animal welfare. Our results can be used as a model to show which habitat requirements an enclosure should fulfil (e.g., coniferous woods for rubbing, wallows, retreats), at what times and in what form food should be given (rooting possibilities), and when the animals should be allowed to rest. These are only a few examples.

The behavioural elements salt ingestion, feeding attempt, getting frightened, stretching, nibbling, wallowing, chasing away, snout knock, and copulation attempt could only be observed at the non-random points and could therefore not be statistically analysed. On the other hand the behavioural elements defecating, urinating, guarding, suckling, scratching one’s bottom and rolling only occurred at the random points. Many important behavioural elements like wallowing and rubbing occur only in certain places and are not necessarily detected by a random distribution. Therefore, it is even more important to observe not only random places but also known whereabouts of the wild boar in order to uncover the entire behavioural repertoire of the species and to describe their needs. In order to increase the chance for documentation of rarely observed behaviours, further studies should be conducted in which more camera traps are placed comparing different localities and populations, as several behaviours could not be observed in this study due to the biotope (habitat) of animals. Furthermore, mating and mating-related fights of males take place from November till January (Meynhardt, 1982; Altmann, 1989; Briedermann, 2009) which is beyond the observation season. The season also has an influence on biotope choice (Keuling, Stier & Roth, 2009). Thus in future studies, it would be advisable to observe wild boar for at least one year via camera traps to get a whole impression of their spatiotemporal behaviour. This year long observation would also account for possible weather influences on the activity and habitat choice of wild boar (Saebel, 2007; Briedermann, 2009; Allwin et al., 2016). Sound recordings could, hence, be taken when looking at courtship interactions to record communication behaviour and to eliminate the influence of the data collection via camera traps.

Currently, the numbers of wolves are rising across Europe (Randi, 2011; Arbieu et al., 2019) and hence, likely influence the behaviour of wild boar, like they do in other species, e.g., roe deer (Bongi et al., 2008) and alpine ibex (Grignolio et al., 2019). This study can serve as a baseline study to record behavioural changes of wild boar in areas in which apex predators are recurring and increasing. To see if the spatiotemporal behaviour changes, future studies could compare different study areas (including or excluding predators, hunting and other human impact, different habitats, different seasons during some consecutive total years).