This study was conducted in the Ashio area of the Ashio-Nikko Mountains (36°54'–36°80'E, 139°22'–139°49'N; Fig. 1). The main study area was approximately 60 km², with an annual precipitation of 2,236 mm and an annual mean temperature of 7.2°C (Tochigi Prefecture 2011). The natural vegetation of the Ashio-Nikko Mountains up to 1,600 m is deciduous broad-leaved forest composed of Quercus crispula Blume, Acer spp., and Fagus crenata Blume. Mixed forests of Tsuga spp. and Betula spp. exist above this altitude. Since 1956, the Japanese government has undertaken major tree planting operations in the area, and today, there are scattered stands of Clethra barbinervis Sieb. et Zucc., Alnus firma Sieb. et Zucc., Robinia pseudoacacia Linn., and Pinus densiflora Sieb. et Zucc.. However, the landscape remains partly open, with a mosaic of grasslands, rough bare land, planted forest, and natural vegetation.

Figure 1. 

Map of the study area, located in the Ashio-Nikko Mountains range in Tochigi and Gunma Prefectures, central Japan. Black lines were trails to observe bears and black circles were the points of vegetation surveyed in the Ashio area.

To clarify the dominant tree species in this area and to select the tree species for analysis, the abundance of tall trees was calculated by a vegetation survey in Ashio study area. Thirty random points were generated in the central Ashio area and set transects (20 × 20 m) in each of the selected points. A handheld GPS receiver (eTrex Legend HCx, Garmin Ltd., Kansas, USA) was used to establish the location where transects were measured. The diameter was measured at chest height of all tall trees and then the total basal area of each tree species was calculated within the transects. The most abundant tall trees in the central Ashio area were R. pseudoacacia, C. barbinervis, A. firma, Betula ermanii Cham., and Q. crispula.

To identify the food items and foraging period for each food item, bears were directly observed feeding during the daytime from April–July in 2013 and 2014. Most bears showed a diurnal feeding pattern during this season (Kozakai et al. 2013) and, therefore, did little foraging at night. We experienced two types of challenges while using direct observation: it was difficult to locate target animals in this temperate broad-leaved forest and observations had to be made without disturbing the animals or changing their behavior. The steep topography and mosaic grasslands of the Ashio study area enabled us to circumvent these problems; we located bears by visually scanning open grasslands and made our observations from the opposite side of the valley using a video camera with a powerful telescopic lens.

To avoid biases toward certain time frames or individuals, we recorded the feeding behavior for as many different bears as was possible. We were able to identify most of the bears by their physical characteristics including body size, chest markings, and the presence of cubs.

The observation frequency was more than once a week (average ± SD: 1.3 ± 0.5 days/week) depending on weather conditions. We walked along trails in the study area and searched for bears using binoculars (Kenko Skymate 8 × 40 mm) as quietly as possible. When bears were located, their behaviors were recorded using a video camera (Panasonic Lumix GH2) with a telescopic lens. In the laboratory, we reviewed the video recordings to identify each food item by tree shape, tree color, or other characteristics, and to calculate the foraging period for each food item. If we could not identify food items from the video recordings, we returned to the feeding location and directly identified the food items based on the signs of feeding. In addition, particularly during 2014, when observation points were proximate to feeding sites and access to accurate feeding location data (via GPS collars fitted on the bears) was available, we download the GPS location data from GPS collars. This allowed us to visit the feeding locations and assess the food items based on signs of feeding such as bear shelves (broken branches on trees made by bears when eating leaves) or claw marks on tree trunks. We defined “ant-feeding” as occurring when the following conditions were met: a bear turned over a stone, put its nose close to the ground, moved its face up and down, moved its front paws, and these actions had to continue for more than 5 s (Fujiwara et al. 2013). During the survey, to protect from attack by bears, we stayed on mountain slopes opposite from the mountain slopes on which the bears were foraging, generally across the valley, and keeping a minimum distance of 200 m between us and bears.

From previous studies, we know that Asiatic black bears mainly consume green vegetation, particularly fresh leaves of woody plants, grasses, and herbs in spring (e.g., Hashimoto and Takatsuki 1997). Therefore, we targeted the leaves of woody plants, grasses, and herbs for nutritional analyses during 2013 samplings.

Six food items were identified by direct observation. For the buds or leaves of all tree species (total of four species), the species could be accurately identified by video data based on confirmation (visiting the feeding site and assessing the feeding signs). However, the species of herbs and grasses could not be identified by video data and only used the data of two species that could be confirmed.

The nutritional contents of tree species consumed by bears were compared to those not consumed by bears. We targeted the leaves of woody plants because a previous study indicated that bears in the study area exclusively fed on tree leaves in spring (Koike et al. 2016). We collected leaves from eight tree species (four consumed species and four non-consumed species) with ten individual trees for each species on two separate occasions (early May and late May). The target non-consumed tree species were dominant species (R. pseudoacacia, C. barbinervis, A. firma, and B. ermanii), selected based on quantitative data. On the other hand, we observed bears feeding on the leaves of four of these species (Malus toringo Sieb., Q. crispula, Salix bakko Kimura, and Elaeagnus umbellata Thund.), and we selected these species as consumed tree species.

Second, to assess the phenologic change in the nutritional value of the three main tree species consumed by bears (M. toringo, S. bakko, and Q. crispula), we collected leaves from ten individual trees of each species once every 2 weeks from leaf flush (early May) to late June (total of four times: early May, late May, early June, and late June). Because feeding on E. umbellata leaves was observed less frequently than feeding on other species, E. umbellata was excluded from this analysis. Third, whenever we confirmed any other plant species being consumed by bears during our 2013 observations, we collected samples from the foraging site as soon as possible.

After collection, samples (10 g (dry weight) for each tree) were kept in paper envelopes and brought to the Tokyo University of Agriculture and Technology where they were dried at 60°C for 48 h. The dried items were milled, placed in plastic tubes, and stored in desiccators until analysis. For each sample, we determined crude protein (CP, % of dry matter), neutral detergent fiber (NDF, % of dry matter), and total energy content (EN, kcal/gdw) according to the methodology prescribed by the Association of Official Analytical Chemists (2003). We determined the nitrogen content (%) using a CN CODER MT-700 analyzer (Yanaco New Science Inc., Kyoto, Japan) and calculated CP using the following formula: CP = nitrogen × 6.25 (Robbins 1993). We analyzed NDF using an Ankom fiber determination system (Ankom²⁰⁰ Fiber Analyzer; Ankom Technology Corp., NY, USA) and analyzed EN using a bomb calorimeter (IKA-Calorimeter C200; IKA-Werke GmbH and Co. KG, Staufen, Germany). All analyses were conducted by the NARO Institute of Livestock and Grassland Science at Tsukuba city, Japan.

After performing nutritional analysis, we used right-angled mixture triangle (RMT; Raubenheimer 2011) analysis to examine the balance of macronutrients and fiber in plant samples. The RMT is a geometric approach used to investigate multidimensional data on the ratios (or balance) of food components in individual foods or food mixtures and is especially relevant to field-based nutritional ecology studies where proportional compositions (as opposed to accurate intake amounts) are the only metric available (Raubenheimer 2011; Raubenheimer et al. 2014). We used a 3-dimensional RMT, where macronutrients were expressed as percentage of total macronutrients (i.e., CP + EN + NDF) on a dry matter basis; CP was shown on the implicit z-axis, the value of which is inversely related with distance from the origin.

We compared the nutritional content of leaves from four tree species not consumed by bears and from four tree species that bears did consume in early and late May by the Bartlett test and used the Kruskal–Wallis test (KW) to determine the differences in nutritional values between tree species. We also compared the nutritional values of M. toringo, S. bakko, and Q. crispula leaves over time and determined the differences in nutritional values related to phenological change by Bartlett test and KW test. We used a single chi-square test to compare the frequency differences from video recordings of bear foraging times for each food item in each half-month period between 2013 and 2014.

We recorded bear foraging behavior for 7 h 25 min during 2013 and 8 h 10 min during 2014 (45 individual and 30 individual foraging behavior events during 2013 and 2014, respectively). The number of observed feeding bouts where we could not identify the bear was 68 and 48 during 2013 and 2014, respectively. The minimum number of identifiable individuals was 10 and nine during 2013 and 2014, respectively. We recorded three bears with a GPS collar in 2013 and two in 2014.

During early April, we did not observe any bear feeding behavior, and the leaf flush had not occurred yet. During late April, the leaf flush had still not occurred, and bears mainly ate Miscanthus sinensis Andersson. Grass (overwintered culms that had stopped growing in November of the preceding year) [74.7% ± 0.5% (mean ± SD) of video-recorded bear foraging time] and S. bakko buds (25.3% ± 0.5%). By early May, the leaf flush had occurred in all tree species except B. ermanii. During this period, bears ate newly emerged leaves of S. bakko (20.7% ± 0.9%), M. toringo (20.3% ± 0.4%), Q. crispula (16.6% ± 2.2%), E. umbellata (16.1% ± 1.6%), Eragrostis curvula (Schrad.) Nees herbs (13.2% ± 2.4%), and unknown grasses (13.1% ± 2.7%). In late May, bears ate the leaves of S. bakko (4.7% ± 0.4%) and Q. crispula (14.7% ± 0.5%) and the flowers and leaves of M. toringo (80.6% ± 0.9%). From June to July, the bears mainly ate ants (96.5% ± 2.1%) and unknown grasses. In early June, bears ate ants (95.0%) and unknown grasses (5.0%) in 2013 and 2014. In late June, bears ate ants (93.9%) and bees (6.1%) in 2013; however, they ate ants (91.5%), unknown grasses (4.5%), and sika deer (Cervus nippon Temminck) carcasses (4.0%) in 2014. In early July, bears ate ants (97.2%) and Cirsium spp. (2.8%) in 2013; however, they ate ants (94.5%) and unknown grasses (5.5%) in 2014. In late July, bears ate ants (98.0%) and unknown grasses (2.0%) in 2013; however, they ate ants (93.5%), unknown grasses (2.5%), and sika deer carcasses (4.0%) in 2014. There were no large differences in leaf flush timing for any plant species between 2013 and 2014 (all <1 week). There were also no differences between 2013 and 2014 frequencies of video-recorded bear foraging times for each food item during each half-month period, except for late June (χ² = 6.586, P < 0.05).

During early and late May, EN of food items had no significant effect on consumption; however, bears were significantly more likely to consume food items with a higher CP (KW test: early May: df = 6, P < 0.05 and late May: df = 7, P < 0.05) and a lower NDF (KW test: early May: df = 6, P < 0.05 and late May: df = 7, P < 0.05). These results indicate that bears use spring plant food items with at least 27% CP and <38% NDF (Table 1).

In CP, Q. crispula leaves decreased from May to late June and was significantly higher in early and late May than in early and late June (KW: df = 3, P < 0.01; Figs 2, 3). In contrast, NDF increased from May to late June and was significantly lower in early and late May than in early and late June (KW: df = 3, P < 0.05). EN was significantly lower during early May and early June than during late June and early July (KW: df = 3, P < 0.01). Bears only consumed the leaves of M. toringo in early May (Fig. 1). CP in M. toringo leaves significantly decreased for each period from early May to late June (KW: df = 3, P < 0.01). NDF was significantly lower during early and late May than during early and late June and early July (KW: df = 3, P < 0.01). EN in M. toringo leaves decreased from early May to early July (KW: df = 3, P < 0.01) and was probably related to the phenology of the leaves. Bears only consumed S. bakko leaves in early and late May (Fig. 1). CP in S. bakko leaves decreased from May to June and was significantly higher in early and late May and early June than in late June (KW: df = 3, P < 0.01). NDF was significantly lower during early and late May than during early and late June and early July (KW: df = 3, P < 0.01). During May and June, EN in S. bakko leaves was not significantly different.

Figure 2. 

Proportion of observed time for each food item and seasonal changes in the nutritional values of Quercus crispula (left) leaves, Malus toringo (center) leaves, and Salix bakko (right) leaves from early May (leaf flash) to late June 2013. A The proportion of time for which bears were observed consuming (2013: black and 2014: gray) B total energy C neutral detergent fiber, and D crude protein. Different lower case letters within each graph indicate significant differences (Kruskal–Wallis test, P < 0.05). White circles indicate when bears were observed consuming Q. crispula leaves (early May and late May), M. toringo leaves (early May), and S. bakko (early May and late May).

Figure 3. 

Right-angled mixture triangles (RMT) depicting the macronutrient balance of Quercus crispula, Salix bakko, Malus toringo, Elaeagnus umbellata, Alnus firma, Clethra barbinervis, Robinia pseudoacacia, and Betula ermanii leaves in early and late May. The RMT on the left a is early May, while the RMT on the right b is late May. Crude protein is represented on the implicit axis which varies inversely with distance from the origin (the dashed gray line indicates 25% protein content).

Other food items were identified and collected for analysis. In late April, bears consumed dead M. sinensis (EN, 4.4 kcal/dgw; NDF, 86.0%; CP, 3.2%) and S. bakko buds (EN, 5.0 kcal/dgw; NDF, 24.8%; CP, 45.0%). In late May, bears consumed M. toringo flowers (EN, 4.2 kcal/dgw; NDF, 28.3%; CP, 28.4%). The herb E. curvula was consumed in early May; however, we were unable to collect samples for analysis.