Exploring the potential of brown bear (Ursus arctos arctos) as a long-distance seed disperser: a pilot study in South-Western Europe

Pyrenean brown bear population

The brown bear population in the Pyrenees mountains decreased regularly throughout the 20th century: 150–200 bears were present in the area at the beginning of 20th century, about 70 bears after the Second World War (Couturier 1954) and finally, only five or six individuals in the western Pyrenees in 1995 with a range totalling about 1000 km² on both sides of the French/Spanish border (Taberlet et al. 1997). In 1993, the European LIFE projects initiated brown bear recovery plans based on reintroductions from Slovenia into the western and central Pyrenees. However, for political reasons, bears were only reintroduced in the central Pyrenees. Two adult females and one male were released in 1996–1997, then another four adult females and one male in 2006 (Quenette et al. 2001, Quenette et al. 2006). After these reintroductions, the population grew and at least 22 individuals were detected in 2012 in two core subpopulations (western and central). These subpopulations are isolated regarding female exchange (Camarra et al. 2014). The western subpopulation has been composed of only two males since 2004 and will certainly disappear if no females are translocated or if they cannot reach the western subpopulation by natural migration.

Study area

The study was conducted in the central Pyrenees mountains, which cover 2000–2500 km² on both the French and Spanish sides of the border. Elevations range from 500 to 3400 m. More than 40% of the area is forested between 800 and 1600 m with pure or mixed stands of beech (Fagus sylvatica) and silver fir (Abies alba). At higher elevations, other deciduous tree species include oak (Quercus spp.), chestnut (Castanea sativa), hazel (Corylus avellana) and common birch (Betula alba). Pasturelands, rhododendron (Rhododendron ferrugineum), heather (Calluna vulgaris) and bilberry (Vaccinium Myrtillus) dominate above 1800 m.

Faeces

Each faeces sample was collected in situ via two main monitoring types: first, through systematic monitoring on predefined study areas and following established protocols (using pedestrian surveys, checking station controls and GPS relocations); then, through opportunistic detection by hikers, hunters, shepherds and members of the Brown Bear Network. Each faeces sample is then validated by the Brown Bear Network.

From a total of 156 faeces samples available and stored for conservation in a freezer, we kept 39 brown bear faeces collected in the Pyrenees from May 1998 to February 2013. Our pool of faeces samples covered the whole active period for brown bears in the Pyrenees (from February until November), with two to four samples per month. In order to be sure that our faeces samples all came from independent defecation events, we screened the samples according to precise detection locality (latitude, longitude), collection date and individual genetical assignment. As much as possible, we selected samples assigned to one of the three GPS-collared bears (Balou, Hvala and Sarousse) introduced from Slovenia in 2006 to reinforce the population, thus corresponding to almost half of our samples (19 faeces). We analysed 5 g of dried matter from each faeces sample. The dried matter was sieved through four different grid sizes (5 mm, 1.5 mm, 800 μm and 100 μm) and all the contents were sorted. We identified six types of macro-elements: hair, ants, bone fragments, honey, plant fragments and seeds. We identified the seeds following the Digital Seed Atlas of the Netherlands (Cappers et al. 2006), the Illustrated Flora of the Pyrenees (Saule 1991) and the reference collections of the National Museum of National History in Paris. We compared the list of species obtained with the list of species inventoried in the study area during botanical surveys and compiled by the National Botanical Conservatory from the Pyrenees and Midi-Pyrenees (CBNPMP, http://cbnpmp.blogspot.fr/).

Plant and seed traits

To address whether bears randomly disperse species or whether a selective dispersal filter based on plant traits (Q2) is at play, we classified the seeds identified in the faeces and the plants present in the bear habitat according to four ecological characteristics:

• Main dispersal mode: We retained the eight main strategies usually taken into consideration (Howe and Smallwood 1982, Kleyer et al. 2008), four abiotic vectors (anemochory, autochory, barochory and hydrochory) and four biotic vectors (dyszoochory, endozoochory, epizoochory and myrmecochory).

  • Fruit type: We distinguished fleshy fruits from dry fruits.

  • Main habitat: We embedded four types (forests; forest edges and clearings; open areas; and wet habitats including bogs and sedge and reed areas).

  • Light requirement: We extracted from plant traits database (Julve 1998) the L-Ellenberg index ranging from one for sciaphilous species to nine for light-demanding species (Ellenberg et al. 1991), and we further classified plants into two groups: mostly sciaphilous to intermediary species (1≤L≤5) as opposed to more light-demanding species (6≤L≤9).

Dispersal distance

To estimate dispersal distance, we used the location data for the three bears introduced from Slovenia in 2006 to reinforce the Pyrenean population (Table 1). Each bear had been (i.e. at the time of introduction) fitted with a GPS/GSM 4000L radio-collar (Lotek Wireless, Fish and Wildlife Monitoring, Newmarket, Ontario, Canada) to provide a scheduled location every 3 h. We kept only the locations with a position dilution of precision, PDOP≤10. For a given season, we chose a location and for each subsequent location, recorded the straight-line distance from the chosen location and the time elapsed. We then calculated the average and maximal distances covered by each bear in spring (post-hibernation), summer and fall (pre-hibernation over-eating) over the cumulative 3-h time slots, following Westcott et al. (2005). All locations were considered to be similar to one another, as we had no information related to the feeding phases of the animals.

We used the median gut retention time (GRT_min, GRT_50% and GRT_max) experimental assessments from Elfström et al. (2013), who measured GRT on captive brown bears fed two contrasted seasonal diets, namely, carcasses or berries (Table 2). We then coupled the distances covered by our three selected individuals with the corresponding median GRT to assess potential seed dispersal distances (Koike et al. 2011).

Statistical analysis

To assess if and how brown bears act as a filter of the local flora through endozoochorous seed dispersal for each plant trait studied, we compared the proportion of plants whose seeds had been found in bear faeces (observed value) with the proportion of plants present in the bear habitat (expected value), for each plant trait category, using χ². We kept a threshold p-value=0.05 for significance. Hence, a non-significant test implies that bears disperse plant species according to their availability in the local flora.

Q1 – Brown bear disperse seeds by endozoochory

Each analysed faecal sample contained from one to five types (n=4) of macro-elements. Plant fragments and seeds were the most frequent types all year round and were, respectively, found in 25 and 31 samples out of a total of 39 samples (Table 3). The other macro-element types are given in decreasing order: ants (n=14), hairs (n=12), honey (n=4) and the least representative were bone fragments (found in only two samples). Seeds were mostly present in summer, fall and winter with a peak in fall, and were rather rare in spring. Ants were more frequent in summer whereas hairs were mostly found in spring.

We sorted out 47 types of seeds, of which 30 were identified to the genus level and 21 to the species level (Table 4), resulting in 16 families and 25 genera. The faecal samples contained on average two to three taxa, but one sample contained eight different taxa.

Q2 – Characteristics of the taxa dispersed and comparison with the taxa present in bear habitat

Seeds from the fleshy fruits of Vaccinium myrtillus or uliginosum, Rubus idaeus, Malus sylvestris and Sorbus sp. but also those from the dry fruits of Thymus sp. and Betula pendula or alba were the most frequently encountered species (Table 4). Among the 30 taxa dispersed and identified, we found 20 grasses or forbs, five shrubs and five trees. Most of the seeds belonged to dry fruits (n=23 vs. n=7 for fleshy ones), but when present in a sample, the seeds from fleshy fruits were much more numerous. A majority of species were light demanding with L-Ellenberg index values between six and nine for 22 taxa. The 30 taxa originated from open habitats (n=11), edges/clearings (n=6), forests (n=9) and wet habitats (n=4). Among them, seven main dispersal strategies were represented: anemochory, autochory, barochory, hydrochory, dyszoochory, endozoochory and epizoochory. Epizoochorous (n=11) and endozoochorous (n=11) taxa were the most numerous.

When we compared the flora dispersed to the flora available in the bear habitat, we found that bear were most likely to disperse endozoochorous and dyszoochorous taxa (χ²₍₇₎=63.34, p<0.0001), and taxa from forest habitats rather than other habitats (χ²₍₃₎=14.52, p=0.002). We found no trend regarding light requirements (L-Ellenberg index, χ²₍₁₎=0.46, p=0.497).

Q3 – Dispersal distances

We summarise in Table 5 the mean and maximal distances covered by male and female brown bears over 18 h. This period of time covers the maximal GRT for bear according to Elfström et al. (2013).

The median GRT (GRT_50% in Table 2) given for a berry-based diet is roughly 6 h. Bears preferentially feed on berries, especially blueberries (Vaccinium sp.) in summer and early fall (Figure 2). Based on these figures and for the three bears we monitored, the median dispersal distances range between 1.09 and 1.34 km in summer and a little less in the fall (between 0.85 and 1.15 km, Table 5).

Figure 2:

Summer (continuous line) and fall (dashed line) average distance covered by Balou ♂ (black diamonds) and Sarousse ♀ (gray triangles) over an 18 h period. This time covers gut retention time (GRT) according to Elfström et al. (2013). For a berry-based diet, median GRT_50% is 5 h 47 min long and for a carcass-based diet, it is 14 h 30 min. Ninety-five percent CI have been deleted for better readability but can be found in Table 5.

For a carcass-based diet, GRT_median is close to 15 h as reported in Table 2. Such a diet is more typical of the post-hibernation behaviour observed in early spring in the Pyrenees, when bears specifically look for carcasses. At this time of the year, the median dispersal distances were longer and ranged between 2.26 and 2.38 km, except for Hvala (Table 5), a female who stayed much longer in the den because she had cubs. As reported in Table 5, for all seasons, maximum dispersal distances ranged from 3 km to more than 10 km, depending on the diet (berry-based vs. carcass-based).