Brown bear communication hubs: patterns and correlates of tree rubbing and pedal marking at a long-term marking site

Study site

The study was conducted in the western half of the Cantabrian Range (NW Spain), a mountain system inhabited by a brown bear population which currently numbers around 230 individuals, with a density of 1.6 individuals/100 km² (Pérez et al., 2014). The study area is located in Fuentes del Narcea, Degaña e Ibias Natural Park (Cangas del Narcea, Asturias). Our study site is located in an area with high quality habitat for brown bears (Naves et al., 2003), including denning and mating areas, areas used by females with cubs, and also vegetation offering plenty of resources used during hyperphagia, when bears feed continuously in preparation for hibernation.

In this area, there are multiple sites used by brown bears for chemical communication. These sites can be easily identified by the presence of a tree, pole or rock that is used for rubbing, often in association with a series of pedal marking tracks leading to the vertical structure that is marked. We selected one site for continuous monitoring on the basis of the evidence of repeated use by brown bears for pedal-marking for more than a decade (Sergiel et al., 2017). As the Cantabrian brown bear population is threatened, we do not provide the exact location of the site due to conservation concerns. The first evidence of ground pedal-marking at this site was obtained in 2002 during an opportunistic observation by one of the authors (DR) of an adult male during the mating season. The site is characterized by an oak tree (Quercus petraea) heavily used by brown bears for rubbing, and by conspicuous marks in the ground made by the bears’ repeated use of the same spots for pedal-marking (a total of 48 marks made by bears’ feet are evident to the human eye).

Sampling protocol

Data were collected by DR at the selected site during long-term monitoring for conservation and management. The Principado de Asturias–Consejería de Agroganadería y Recursos Autóctonos granted data access, and DR was authorised to participate by exp-no. 2016/033072, Principado de Asturias-Consejería de Hacienda y sector Público. An automatic camera trap (Bushnell Trophy digital camera trap #19466 with motion triggered day/night recording) was set up between January 2012 and January 2016, during which time it was working almost continuously. Initially, between January 2012 and April 2012, the device was placed laterally in a low position from which the tree marked by brown bears was visible. Data obtained during these first four months were not used in the analyses. After this initial sampling, the camera trap was mounted in a zenith position (directly above the site) at a height of six meters on the main trunk of the marked tree to obtain a standardized field of view and to reduce direct interference with bears and other animals. The field of view of the camera trap covered an area of about 100 m². The camera trap was programmed to shoot one-minute videos, with a 10-second interval between consecutive videos. We considered a visit event as the group of videos that are not more than 20 min apart. This time window was selected following visual inspection of the plot of the cumulative proportion of videos sorted by the time to the next video (Fig. S1). For comparative purposes, we also used a 20-minute time interval to define visit events for other species. Note that a visit can include more than one individual bear, as occurs in the case of females with cubs or males and females moving together during the mating season. The weather data were obtained from the nearby automatic station of Leitariegos, belonging to the Spanish Agencia Estatal de Meteorología (AEMET).

Individuals and communication behaviors

In the Cantabrian Mountains, the steep slopes and low forest cover make it relatively easy to observe brown bears, especially during spring and summer. Individuals present in valleys are detected by scanning the area with spotting scopes from vantage points. This method is used to obtain annual counts of the number of females with cubs of the year and as a long-term method to census this population (Wiegand et al., 1998). As a result, some of the individuals moving in the study area are known, especially when they have some identifying marks, and are thus easily distinguished from other individuals. The professional technicians doing those censuses are experts in recognizing the sex and age of individuals by specific traits under good observation conditions. We classified the recorded individuals into the following sex and age categories: (1) adult males, identified by the combination of large size, and neck and head shape; (2) adult females, when accompanied by cubs, or identified by their size, head and neck shapes, and explicit behavior in the presence of other individuals, often adult males in the mating season; (3) cubs, individuals in their first year or in their second year until May and always accompanied by their mother; (4) juveniles, independent individuals in their second year of life from June onwards and in their third year, clearly smaller in size than adults and usually accompanied by siblings; and, (5) undetermined sex and age class, which included the remaining individuals.

In the case of adult males, some individuals were identified by comparison with known animals observed in repeated sightings at other sites in the study area. These individuals were characterized by a combination of body size, head shape, coat color patterns and especially the very characteristic light-colored permanent markings, normally present on their necks (see description of individualized bears in Supplemental Information). The Cantabrian brown bear population is characterized by its small size and the large variability shown by individuals in coat color and the common presence of markings especially on their necks (Clevenger & Purroy, 1991). In other cases, we were able to temporarily classify some individuals in an age and sex class or even identify them during shorter periods of time because they were associated with other bears in seasonal or yearly groups such as mating pairs, females with dependent cubs, and groups of independent juveniles repeatedly seen in the area. Females are more difficult to identify individually on a permanent basis. We used the number of accompanying cubs to establish a minimum number of females visiting the site each year. We did not attempt to identify other types of individuals such as independent juveniles and cubs.

We classified the behaviors displayed by brown bears in the videos into the following types: (1) sniffing pedal marks, when an individual stops or slows its pace and puts its nose to the pedal marks on the ground; (2) pedal-marking, performed by a walking bear with the particular gait of twisting its fore and hind feet on the ground in specific depressions repeatedly used by that individual and other bears during previous visits; (3) tree-sniffing, when an individual calmly puts its nose to the trunk of the rubbing tree; (4) tree-rubbing, when a bear vigorously rubs its back, neck or shoulders against the trunk of the tree while standing on its hind legs; and, (5) other behaviors, in which a bear usually walks in and out of the field of vision. In the videos recorded at the study site we did not detect any clear instance of scratching the tree (clawing; Taylor, Allen & Gunther, 2015). For each visit event we determined if each type of behavior was performed (presence/absence of the behavior, not the number of times) by each bear in the available sequence of videos.

Analyses

First, we described the overall use of the site and the behaviors performed by the visiting brown bears over time and by age and sex classes. Then, we analyzed which variables were associated with the observed patterns (Table 1). We hypothesized that the probability that bears visited the marking site in a given day and performed one of the behaviors in each visit was affected by not only the time elapsed since the previous visit by a bear, but also the season, distinguishing between mating season (April, May and June) and non-mating season (other months), as well as the age and sex class of the focal individuals, and, in the analyses where it made biological sense, by the weather conditions that occurred between visit events affecting the duration of the chemical signals. We performed Generalized Linear Mixed Models (GLMMs) on the response variables (occurrence of the specific behaviors) using a binomial error distribution and year as a random factor. As some individuals were repeatedly observed, there could be some pseudoreplication problem. Solving this issue is not easy as a fraction of the observations correspond to unknown animals. Nevertheless, and in order to check if pseudoreplication was an issue, we repeated the analyses of the selected models adding individual ID as an additional random factor (unidentified individuals were grouped under a single individual label; results are shown in the Supplemental Information). Models were run with the potential combination of biologically meaningful explanatory variables within each group of response variables (Table 1). To reduce the effect of multicolinearity, when two predictors were correlated, we selected the one with a stronger association with the dependent variable. From the resulting models, we report and interpret only those within ΔAIC < 2. We computed the marginal and conditional R² for each selected model (Nakagawa & Schielzeth, 2013). For inference we used a cut-off level of P < 0.10. Analyses were performed in R vs3.3.3 with lme4 (v1.1-19; Bates et al., 2015) and MuMIn (v1.43.6; Bartoń, 2019) packages.

Communication behaviors

The typical sequence of a visit consists of a bear approaching the tree following the path where it can sniff the depressions in which animals pedal mark, performing pedal-marking itself, stopping at the tree, sniffing it, and, finally, rubbing against the trunk (see Video S1). This sequence can vary with different combinations of behaviors and in different orders, and some parts of the sequence can be repeated. On one occasion, a male also rubbed its body against pedal marks. There was no apparent communication behavior in 22% of the visits, although some could have occurred out of the field of view of the camera trap.

Out of a total of 482 recorded behaviors, the majority corresponded with some form of chemical communication (87%). Communication behaviors occurred in most months except January and February (hibernation period; Table S4). Sniffing of pedal marks was less frequent (58, 12%) than pedal-marking (113, 23%); while tree-sniffing (153 cases, 31%) was more frequent than tree-rubbing (96, 20%; Table S4).

The communication behaviors displayed by bears varied greatly among age and sex classes. All sex and age classes performed pedal- and tree-sniffing. Individuals identified as adult males performed most of the pedal-marking (107 cases, 95%) and, to a lesser extent, tree-rubbing (63 cases, 66%, Fig. 2). Interestingly, adult females did not perform pedal-marking, while juveniles did so at very low frequency (Fig. 2). Tree-rubbing was performed by all age and sex classes, but at higher frequencies by males (Fig. 2).

Males and females sniffed the pedal marks in 23% and 21% of their visits, respectively; while cubs, juveniles and undetermined bears did so in 61%, 48% and 26% of their visits, respectively. The probability that a bear sniffed the pedal marks during a visit was higher outside the mating season (Table 3). Also, the lower the average precipitation and the average temperature in the preceding days, the higher the probability of sniffing the pedal marks (Table 3, Tables S3 and S8). Finally, the probability of sniffing the pedal marks tended to be negatively related to the time elapsed since the last time a bear performed pedal-marking at the site (Table 3; Tables S3 and S8).

Males performed pedal-marking in 81% of their visits to the site. They both pedal-marked and sniffed the pedal marks in 20% of their visits. Juveniles and undetermined bears performed pedal-marking in 17% and 7% of their visits, respectively, while females and cubs never pedal marked. The probability of performing pedal-marking by male bears visiting the site was positively associated with tree-rubbing by the same individual and negatively with the time elapsed since the previous visit of a bear that pedal-marked at the site (the shorter the time, the higher the probability of pedal-marking, Table 3; Tables S3 and S8). The association of pedal-marking probability with the remaining factors was weaker (Table S3).

Males sniffed the tree in 59% of their visits, while adult females did so in 51% of their visits. Cubs, juveniles, and undetermined individuals showed interest in the tree, sniffing it in 61%, 48%, and 26% of their visits, respectively. Interestingly, the probability of sniffing the tree by a visiting bear was higher the longer the time elapsed since the previous tree-marking event and negatively related to the precipitation during that period (Table 3; Tables S3 and S8), and was not affected by the sex or age class of the individual. Nevertheless, the model was not very explanatory (Table 3).

Males performed tree-rubbing in 48% of their visits. They engaged in both pedal-marking and tree-rubbing during the same visit on 43% of their visits and tree-rubbing and tree-sniffing in 35% of their visits. Adult females rubbed against the tree in just 9% of their visits. Juveniles, cubs and undetermined individuals tree-rubbed on 39%, 34% and 14% of occasions, respectively. Adult males and juveniles had higher probabilities of tree-rubbing during their visits than females (Table 3; Tables S3 and S8). The probability that a bear performed tree-rubbing during a visit was positively associated with tree-sniffing and pedal-marking by the same individual (Table 3; Tables S3 and S8), and tended to be positively associated with the time since the previous tree-rubbing event (Table 3; Tables S3 and S8).

Several recognizable individuals visited the site repeatedly (Supplemental Information), some of them throughout the study period. Four adult males visited the site between 10 and 35 times during the study, with up to 15 visits in one year (M1 to M4, Table S6). These males were frequent markers; for example, M2 and M3 were responsible for most of the instances of pedal-marking (59%, Table S7), while M2 was the bear that most frequently displayed tree-rubbing behavior (43%, Table S7). Additionally, other males visited the site sporadically (Table S9). These additional males were known individuals that were repeatedly observed near the study site (at least four additional males in 2012, five in 2013 and 2015, and seven in 2014). A minimum of one female visited the site in 2013 and 2015, two in 2014 and three in 2012. The minimum number of different individual bears visiting the site per year ranged between 11 in 2013 and 18 in 2015 (Table S9).

Sending and receiving information

The importance of chemical communication at the site varied as a function of the individuals, depending on their sex, age, and presumably other conditions such as dominance or breeding status. Nearly half of the visits to the marking site were made by animals identified as adult males. They were responsible for most pedal-marking, and, to a lesser extent, tree-rubbing behaviors. Both behaviors were strongly associated when performed by adult males. Some males visited the site very often while others were more sporadic. Interestingly, some males marked in most of their visits while others mostly acted as information receivers. This may reflect a structure of dominance in the males sharing the area. Females, on the other hand, never pedal-marked and rarely rubbed the tree, and neither did the cubs accompanying their mothers. Young animals (of unknown sex) showed an intermediate pattern between males and females. Tree-rubbing was more frequently displayed by bears which also sniffed the tree and performed pedal-marking and positively related with the time elapsed since a previous tree-rubbing event, typically describing the behavioral sequence of visiting males. Male brown bears have seasonally enlarged sebaceous glands on their back and prominent eccrine, apocrine and sebaceous glands in their feet; glands that are more active during the mating season, in association with their increased testosterone levels (Sergiel et al., 2017; Tomiyasu et al., 2018). Therefore, males acted as main sources of chemical messages at the site, as has been shown in other study areas (Clapham et al., 2014; Lamb et al., 2017).

Sniffing behavior, especially that of ground marks, is less obvious and therefore more likely to go unnoticed in videos. Nevertheless, all types of individuals showed interest in the chemical marks, acting as genuine information receivers. The probability of sniffing the marks during a visit was affected by weather conditions, with higher temperatures and precipitation in the preceding days reducing the probability of sniffing ground marks, a pattern that was not associated with actual pedal-marking, and higher precipitation negatively affecting tree-sniffing. The diluting effects of precipitation and temperature on the volatility of the odorous molecules left by bears at the marking site are a possible interpretation of these results. Interestingly, the probability of sniffing the tree was higher the longer the time elapsed since the previous visit, while it was the opposite for ground sniffing, suggesting a differential detectability between the chemical compounds secreted by pedal and back glands and among different substrates.

Why brown bears visit these sites

Brown bears use chemical marking to convey information from senders to receivers. Why they do this and what type of information is transferred is still a matter of discussion. The chemical profiles of pedal and shoulder secretions indicate that they contain information on at least the sex and reproductive status of the individual (Sergiel et al., 2017; Tomiyasu et al., 2018). Additionally, it would not be surprising if information on the actual individual is also provided, as seems to occur with secretions from anal sacs (Rosell et al., 2011; Jojola et al., 2012). In species that normally exhibit a solitary non-territorial use of space, knowing the individuals whom they may encounter is quite valuable. Several non-exclusive hypotheses have been proposed to explain scent-marking in brown bears: self-advertisement for mate attraction, communication of individual dominance, competitor assessment and infanticide avoidance, with different roles depending on bear density (Clapham et al., 2012; Lamb et al., 2017). Our results show that chemical communication in brown bears is complex. Males are the main senders and also the main receivers, with some of them marking a lot while others tend to mostly receive information, indicating communication of individual dominance and the ability to assess male competitors. Male bears mark all year round but with a main peak during the mating season, a period of intense competition. This pattern has also been found at rubbing trees, both natural and artificially created to collect bear hairs (i.e., tree hair traps), in different ecosystems (Green & Mattson, 2003, Karamanlidis et al., 2010; Sato et al., 2014; Berezowska-Cnota et al., 2017; Lamb et al., 2017).

Females seem to visit the site less often, but all year round, and when they do, they are especially interested in receiving information. Knowing which males are moving around and their social dominance is very important for females in mate selection, since mating with the more dominant males that are present all year round would minimize the overall risk of infanticide to their litters. Additionally, females with cubs of the year may benefit from knowing if a new male enters the area (Bellemain et al., 2006). Although more rarely, females, juveniles and cubs also rub trees, but it is unclear why they do it. In the case of juveniles learning by imitation may be the main reason (Clapham et al., 2014). Given that the sebaceous secretion in the shoulder of males is linked to testosterone levels, the secretion of females, cubs and juveniles can be expected to be testimonial or simply non-existent. If that is the case, their tree-rubbing may serve the purpose of masking their odor with that of adult males roaming the area. The resulting increase in chemical similarity could help to reduce the risk of infanticide by scent-matching (Gosling & McKay, 1990). If this interpretation is correct, tree-rubbing would have a scent-marking purpose only for males, while helping females and cubs to obtain a chemical camouflage by scent-rubbing as well as transitionally being part of the learning process of juveniles. In summary, there is no single best hypothesis to explain the role of these communication hubs, with the most plausible being a complex combination of dominance, mate selection, competitor assessment, mate selection and infanticide avoidance.

Brown bear communication hubs

Undoubtedly, sites like the one we monitored are important for brown bears at the population level. Our results show that the tree and the trails leading to it form a communication hub that most bears living in the area use to share and obtain information. Bears were the most frequent visitors to our site despite the easy accessibility and the fact that bears are not the most common large mammal. Bears choose specific trees in places that are well situated for the passage of other individuals (Green & Mattson, 2003; Sato et al., 2014). At these sites there is an association between different communication behaviors, with marking behaviors triggering the subsequent sniffing and marking of later visitors (Berezowska-Cnota et al., 2017). Nevertheless, these sites are not uncommon. In the vicinity of our site there were other trees used repeatedly by bears for marking (see Video S1). Brown bears maintain a dense system of marking sites that allow for a complex communication network over large spatial scales. Although they are not easy for humans to locate, several authors report varying densities of marking sites depending on bear density, including 0.26 sites/km² in the Italian Alps, 0.4 sites/km² in Hokkaido, Japan, 1.4 sites/km² in the Russian Komi Republic, 20 sites/km² in British Columbia, and 27 sites/km² in the Valley of Geysers on Kamchatka Peninsula (Lloyd, 1979; Sato et al., 2014; Seryodkin, 2014; Tattoni et al., 2015). Many of these studies describe trails evidencing pedal-marking (e.g., Clapham et al., 2013; Seryodkin, 2014). There are open questions that remain to be answered, such as the heterogeneity in the use of the multiple marking sites available to brown bears within their home ranges or the variability in marking intensity within and across populations.