Knowing animals' gut retention time (GRT) for important food items is critical when using non-invasive studies based on faecal remains, e.g. when analysing nutritive quality of food, or relating diet or behaviour to movements. We analysed GRT in six captive brown bears Ursus arctos, after feeding on either berries (a mixture of bilberry Vaccinium myrtillus and lingonberry V. vitis-idaea) or animal carcasses (either reindeer Rangifer tarandus, European rabbit Oryctolagus cuniculus, domestic pig Sus scrofa domestica, cattle Bos taurus or horse Equus ferus caballus). Median GRT50% (i.e. when 50% of all faeces containing experimental food had been defecated) was 5 hours and 47 minutes (1st and 3rd quartiles = 4 hours and 36 minutes and 7 hours and 3 minutes; N = 20) after feeding on berries and 14 hours and 30 minutes (1st and 3rd quartiles = 10 hours and 9 minutes and 16 hours and 57 minutes; N = 20) after feeding on carcasses. Median GRTmin (i.e. first defecation comprised of experimental food) was 3 hours and 5 minutes (1st and 3rd quartiles = 1 hour and 51 minutes and 4 hours and 12 minutes; N = 21) for berries and 8 hours and 2 minutes (1st and 3rd quartiles = 6 hours and 14 minutes and 10 hours and 44 minutes; N = 20) for carcasses. Median GRTmax (i.e. last defecation comprised of experimental food) was 15 hours and 27 minutes (1st and 3rd quartiles = 11 hours and 36 minutes and 17 hours and 16 minutes; N = 21) for berries and 16 hours and 16 minutes (1st and 3rd quartiles = 12 hours and 11 minutes and 17 hours and 27 minutes; N = 20) for carcasses. A carcass diet had 6 hours and 26 minutes ± 1 hour and 56 minutes (SE) longer GRT50% than a berry diet (N = 39), despite low variation in food intake. Activity level, feeding time (midday/midnight), sex, age (subadult/adult), ingested amounts of food, prior food remains processed by the gut (i.e. cumulative faeces weight) and defecation rate did not influence the GRT50%. Our reported GRT estimates are reliable values to be used within research and management to relate diet based on faecal remains to habitat use for common and important food items used by Scandinavian brown bears.
Analyses of faecal remains allow non-invasive studies of, for instance, a species' distribution, habitat use and diet (Putman 1984). Data based on faecal remains can also be related to animal movements (e.g. using GPS positions). However, these studies often require knowledge of the time for ingested food to pass through the digestive tract, hereafter called gut retention time (GRT). Therefore, knowledge of the GRT is valuable for many types of studies. Knowing the GRT allows sampling faeces within a defined time frame, thus uniting feeding patterns with spatio-temporal data of individuals. In domestic pigs Sus scrofa domestica, a short GRT is associated with increasing proportions of ingested fibre (Partanen et al. 2007). When problem carnivores are shot (e.g. because of depredation incidents), the GRT also defines the time frame during which depredated foods are expected to be found in the digestive tract of destroyed animals. The GRT has been studied for several aquatic and terrestrial carnivores (Edwards et al. 2001, Hall-Aspland et al. 2011), as well as for omnivores (Tsuji et al. 2011). In bears, Ursidae, the GRT has been studied in omnivores with mainly a vegetative diet (giant pandas Ailuropoda melanoleuca; Dierenfeld et al. 1982), almost exclusively carnivores (polar bears Ursus maritimus; Best 1985), and omnivores with relatively large variation in their diet (Asiatic black bears Ursus thibetanus; Koike et al. 2010, and brown bears Ursus arctos; Pritchard & Robbins 1990). Pritchard & Robbins (1990) estimated the GRT for hair when feeding on carcasses and (chromium-marked) clover Trifolium repens in North American brown bears confined in small cages.
We analysed the GRT of captive Scandinavian brown bears for two common food items; berries and meat from carcasses. Meat (i.e. newborn calves of reindeer Rangifer tarandus and moose Alces alces) and berries (i.e. bilberry Vaccinium myrtillus, crowberry Empetrum nigrum and lingonberry Vaccinium vitis-idaea) constitute important foods for Scandinavian brown bears during spring and autumn, respectively (Dahle et al. 1998, Persson et al. 2001). Thus, although bears may have a mixed diet, they often feed on animals and berries during separate periods of the year. We compared GRT in relation to sex and age classes of bears, feeding time (midday/midnight), activity, diet (berry/carcass), weights of experimental food and supplemental food, cumulative weight of faeces and defecation rate. We hypothesised a longer GRT for a carcass than a berry diet, because Pritchard & Robbins (1990) reported higher digestibility (i.e. less material to be processed by the gut) and lower fibre content for meat than for berries.
Material and method
We studied GRT on six captive animals, three females (two subadults 2.5-year old and one adult 8.5-year old) and three males (two subadults 3.5-year old and one adult 10.5-year old) in the Orsa Bear Park, Sweden, during August of 2010. For none of the animals there were earlier reports or indications of gastrointestinal diseases. The four subadult bears were kept together in the same enclosure, whereas the two adults were kept together in a separate enclosure, both encompassing approximately 10,000 m2. We used two individuals per experiment, and consecutive experiments were separated by a minimum of 48 hours for each individual.
During an experiment, bears were confined to an enclosure encompassing 400 m2 for 24 hours. In order to standardise and improve the detection of experimental foods, individuals were given no food, except for ca 200 g of dog food pellets provided immediately after entering the experimental enclosure, and after four hours they were given the experimental foods. Bears were either given their experimental foods at midday (12:00) or at midnight (00:00) in order to control for last routine feeding between experiments and diel behaviour (Moe et al. 2007), which may affect gastrointestinal functions (Bron & Furness 2009). Between the experiments, bears were fed fruits (i.e. grapes Vitis vinifera, apples Malus sp. and oranges Citrus sp.) and bread daily at 14:00. Thus, ingestion of experimental food at midday took place ca 22 hours and midnight feeding ca 10 hours after the last routine feeding. Bears had access to carcasses (parts of domestic pig, cattle Bos taurus or horse Equus ferus caballus) within 12 hours before two experiments on carcass diet and one experiment on berry diet. Bears always had access to water and were given corn Zea mays weighed as fed after the experimental food had been consumed.
We mixed the experimental foods with 50-100 g plastic beads (of 5 mm in diameter), which functioned as solid markers to assist detection of experimental food items in the faeces and to confirm that faecal remains were derived from experimental foods. Experimental feeding of berries was comprised of a mixture of bilberry and lingonberry with a large proportion of the former. Experimental feeding of carcasses was comprised of meat, bones and fur from either bear-killed domestic reindeer calves, domestic rabbits Oryctolagus cuniculus or parts of domestic pigs, cattle or horses. All provided experimental food was consumed except larger pieces of bones or fur, which were subtracted from the weight as fed after the experiments. We only used data from experiments in which bears finished consuming the experimental food within two hours.
We video-recorded the animals during each experiment, using light-equipped cameras and recording capability within infrared wavelengths, noted start and end time of feeding, time of defecation, measured with an accuracy ± 1 second and scored an activity level every 10 minutes as active (standing/moving) or passive (laying/sitting down). After the bears had been released back into their main enclosures, all faeces were collected, labelled in order to relate each faeces to the time of defecation and bear identity (based on the video-recordings) and immediately weighed on an electronic scale. We examined faecal remains to detect the presence of markers, and we separated faeces containing experimental food item, i.e. berry or carcass, from those containing only corn. No corn defecation occurred before the first defecation containing remains of ingested experimental food, and the last defecation during experiments contained corn. The mean ± SD air temperature during the experiments was 15°C ± 4 (SD) at 12:00 and 12°C ± 3 (SD) at 00:00. Our study was approved by the Ethical Committee of Animal Research in Umeå, Sweden (permit D nr A 75-10).
We used both start and end times of feeding to calculate GRT. GRTmin denotes the time elapsed before the first defecation containing experimental food with markers, and GRTmax denotes the time elapsed before the last defecation containing experimental food with markers, after the ingestion of experimental food. GRT50% denotes the time when 50% of the cumulative weight of faecal remains of experimental food with markers had been defecated after the ingestion of experimental food. We report median values, because distributions were non-normal, and to avoid overestimating the GRT (Ormseth & Ben-David 2000).
We estimated GRTmin, GRTmax and GRT50% using only faeces with confirmed presence of experimental foods and markers. However, the total amount of material processed in the gut during and after the time of ingesting the experimental food affects the available volume in the gut before the next defecation (e.g. the GRT50%). Thus, material processed by the gut, measured as defecation rate and cumulative weight of defecations after ingesting experimental food prior to the GRT50%, may affect GRT50%. Therefore, when calculating defecation rate and cumulative weight of defecations prior to the GRT50% defecation, we included all defecations (i.e. also faecal remains of corn and not containing experimental food or markers). We used linear mixed models (LMM) to analyse GRT50% in relation to the following fixed factors: sex, age class (adult/subadult), activity score (% active), feeding time (midday/midnight), diet (berry/carcass), weight of ingested experimental food, weight of ingested supplemental food (corn), cumulative weight of defecations prior to GRT50% and defecation rate prior to GRT50%. We included an interaction term between sex and age classes to control for potential differences among these classes, because bears are sexually dimorphic (Rode et al. 2006) and body size has been suggested to influence the GRT among herbivores (Demment & Van Soest 1985). We calculated the defecation rate prior to the GRT50% as the number of defecations divided by the period from midpoint of feeding (between start and end times) until the GRT50% defecation occurred. We calculated GRT50% using the midpoint between the start and end of feeding in our LMM. We used bear identity and experiment as random effects.
We constructed a candidate set of 14 LMMs a priori and selected the most parsimonious LMM based on Akaike's Information Criteria scores for small sample sizes (AICc) and AICc weights (AICcw; Akaike 1973, Burnham & Anderson 2002). We used the ‘lme4’ package (Bates & Maechler 2010) for statistical modelling and generated β and its 95% highest posterior density interval (HPD) for the fixed effects of the regression models with a Markov Chain Monte Carlo algorithm (MCMC) using 1,000 simulations, package ‘LMERConvenienceFunctions’ (Tremblay 2011) in R.2.14.1 (R Development Core Team 2009). We considered effects significant when the HPD 95% around β did not include 0. Outliers were controlled for by using Cleveland dotplots and multicollinearity by using variance inflation factors (Zuur et al. 2009). The number of observations (N) deviated among analyses, because the factor activity score was missing for one animal during one experiment, and weights of faeces were missing for one animal during another experiment.
Gut retention times and defecation rates for berries and carcasses
Median GRT50% from the midpoint time of feeding was 5 hours and 47 minutes (1st and 3rd quartiles = 4 hours and 36 minutes and 7 hours and 3 minutes; N = 20) for berry diet and 14 hours and 30 minutes (1st and 3rd quartiles = 10 hours and 9 minutes and 16 hours and 57 minutes; N = 20) for carcass diet (Fig. 1). Median GRTmin from the midpoint time of feeding was 3 hours and 5 minutes (1st and 3rd quartiles = 1 hour and 51 minutes and 4 hours and 12 minutes; N = 21) for berry diet and 8 hours and 2 minutes (1st and 3rd quartiles = 6 hours and 14 minutes and 10 hours and 44 minutes; N = 20) for carcass diet. Median GRTmax from the midpoint time of feeding was 15 hours and 27 minutes (1st and 3rd quartiles = 11 hours and 36 minutes and 17 hours and 16 minutes; N = 21) for berry diet and 16 hours and 16 minutes (1st and 3rd quartiles = 12 hours and 11 minutes and 17 hours and 27 minutes; N = 20) for carcass diet. Descriptive estimates of GRTmin, GRTmax and GRT50% are shown in relation to start and end points of feeding experimental foods in Table 1.
Median (1st and 3rd quartiles) defecation rate between start of feeding and last defecation of experimental food among experiments and individuals was 7.1 defecations/24 hours (6.1 and 9.4; N = 21) for berry feedings and 4.0 defecations/24 hours (1.2 and 5.4; N = 20) for carcass feedings.
Effects of diet, activity, sex, age, weights of food and faeces and defecation rate on GRT50%
The most parsimonious LMM included sex, age, (adult/subadult), feeding time (midday/midnight), activity score, prior defecation rate, diet (berry/carcass) and the interaction between sex and age class based on AICc (ΔAICc = 0.00 and AICcw = 0.97; Table 2). Thus, this LMM excluded weight of experimental foods, supplemental food and prior cumulative faeces weight (see Table 2). Diet (berry or carcass) was the only fixed factor with a HPD 95% interval around βMCMC that did not include 0; it had a β/SE of 3.3. A carcass diet had 6 hours and 26 minutes ± 1 hour and 56 minutes (SE) longer GRT50% than berries (N = 39 feedings; Table 3). All other fixed factors in this model had HPD 95% intervals around βMCMC that included 0, and β/SE ratios were ≤ 2.0; i.e. sex had β/SE = 2.0, age (adult/subadult) had β/SE = 1.0, feeding time (midday/midnight) had β/SE = 0.7, activity score had β/SE = 0.5, prior defecation rate had β/SE = 0.1 and interaction between sex and age class had β/SE = 1.3 (N = 39 feedings; see Table 3). Descriptive estimates for continuous fixed factors used in our LMM data set are shown in Table 4.
We found no relationship between GRT50% and activity levels of the animals, which were constrained within ca 400 m2 during the experiments. Our reported median GRT50% of 14 hours and 30 minutes after feeding on carcasses is similar to a mean GRT for hair ingested by North American brown bears when feeding on carcass, based on amount of digested marked and unmarked hairs per defecation, 13 hours ± 2 hours (SD; Pritchard & Robbins 1990). However, Pritchard & Robbins (1990) used animals constrained within cages with a maximum dimension of 2.4 m. This suggests that GRT is not related to activity levels, and that our estimates of GRT50% are reliable values to use within research (e.g. to compare diet based on faecal remains with movements) and management (e.g. for how long to expect to find ingested livestock in the gut of shot bears) for common food items used by brown bears.
We found no relation between GRT50% and feeding time (midnight or midday). The main sleeping period of the bears used in this experiment was between midnight and sunrise, whereas the last routine feeding took place 22 hours before experimental feeding at midday and 10 hours before experimental feeding at midnight. This suggests that there is no effect from a circadian activity pattern on GRT50% and/or effects from last ingestion before our experiments of GRT50%.
We provided the same amount of food during all experiments, and this may explain why we found no relationship between food intake (i.e. weight of ingested experimental or supplemental food) and GRT50%, as well as no relationship between cumulative faecal weights or defecation rate prior to the GRT50% defecation and GRT50%. In the carnivorous leopard seal Hydrurga leptonyx, Trumble et al. (2003) reported similar GRT among experiments with different feeding frequency. However, it is possible that a larger variation in food intake would have had a larger effect on the GRT in our study, because larger food intake may shorten the GRT due to gut volume constraints. A negative correlation between food intake and GRT has been reported in omnivores, e.g. mice (McClelland et al. 1999) and herbivores (Clauss et al. 2007).
We found no differences in GRT50% between subadults and adults, nor between female and male bears (i.e. groups with smaller and larger body sizes), even though foraging efficiency may decrease with increasing body size in bears (Welch et al. 1997, Rode et al. 2001). GRT does not change with body size in dogs Canis familiaris (Boillat et al. 2010) and primates (Lambert 1998). Steuer et al. (2011) concluded that body mass alone poorly explained differences in GRT between small and large herbivorous ungulates.
The GRT50% for the carcass diet was 6 hours and 26 minutes ± 1 hour and 56 minutes (SE) longer than of berries, despite low variation in food intake. Pritchard & Robbins (1990) reported higher digestibility for carcasses (93%) than for blueberries Vaccinium corymbosum (64%). A higher digestibility of carcasses compared to berries results in a reduced amount of faecal remains and, hence, the gut can contain more faecal remains before the gut volume is filled and defecation occurs after feeding on carcass. Giant pandas feeding on bamboo Phyllostachys aureosulcata, which is rich in fibre content, have short GRT, probably because they ingest large amounts of poorly digestible food (Dierenfeld et al. 1982). Partanen et al. (2007) reported a shorter GRT and lower digestibility with increasing proportions of ingested fibre for pigs that were fed the same amounts of food. In humans, ingested fibre is known to improve bowel movements and produce softer faeces (Klosterbuer et al. 2011) and fresh bilberry has been described as having a laxative function (Jaric et al. 2007). In birds, ingested seeds have been suggested to have a chemical laxative function by shortening the GRT (Murray et al. 1994). The dietary fibre content of berries is five times higher than in carcasses (Pritchard & Robbins 1990). Thus, the much shorter GRT50% for berries compared to carcasses in our study may be a result of lower digestibility in combination with increased gastrointestinal activity after ingesting berries.
Berries constitute the most important food item for Scandinavian bears during hyperphagia in summer and autumn (Dahle et al. 1998). The potential median and maximum seed dispersal distances for berries, based on our GRT50% and GRTmax for berries, are 4.2 km and 11.1 km, respectively, when combining our results with a reported median rate of movement of 0.72 km/hour by Scandinavian brown bears (Moe et al. 2007). We found that the median (1st and 3rd quartiles) defecation rate of 7.1 (6.1 and 9.4) defecations/24 hours when bears foraged on berries is similar to the 7.2 defecations/day during autumn reported by Roth (1980) in captive brown bears fed a diet of mostly plants.
Our results suggest that GRT estimates are reliable to use in research where GRT after berry/carcass diet of bears are required. When combined with positioning data, GRT constitutes an important tool for determining where food remains found in faeces have been consumed and where remains of consumed foods will be excreted. The GRT also constitutes an important tool for management by defining a time frame in which to expect finding particular food remains in bears, e.g. livestock remains.
We thank our reviewer M. Clauss for comments on earlier drafts. We thank the Orsa Bear Park, Grönklitt, Sweden, and their staff, for providing us the opportunity to test GRT on their bears. We are most grateful to Olle Svensson AB, a part of the Nordic Food Group, for providing wild berries for the feeding experiments and Munkplast AB for providing plastic beads to use as experimental markers. We also thank the Swedish Environmental Protection Agency for financial support. This is scientific publication number 150 from the Scandinavian Brown Bear Research Project.
Gut retention time (GRT; in hours:minutes) in six captive Scandinavian brown bears (two 2-year-old and one 8-year-old female; two 3-year-old and one 10-year-old male), after feeding on either berries or carcass (meat with bones and fur), at the Orsa Bear Park, Sweden, during 2010. GRT is combined between feeding at midday and midnight. Maximum time elapsed between start and end times of feeding was 1 hour and 39 minutes. GRTmin and GRTmax are based on 21 berry feedings and 20 carcass feedings, and GRT50% is based on 20 berry feedings and 20 carcass feedings.
Model selection based on AICc values (wi = AICc weights), finding the most parsimonious linear mixed model when fitting gut retention time when 50% of faeces with experimental foods had been defecated (GRT50%) for six captive Swedish brown bears, with bear ID and experiment as random effects, using a priori set of 14 candidate models. A variable on grey background represents its exclusion. A = subadult or adult, Ac = Activity score, CB = carcass or berry diet, DN = midday or midnight feeding, Dr = defecation rate, S = sex, Wc = cumulative weight of faeces , We = weight of experimental food, Ws = weight of supplemental food and * = an interaction between two factors.
Factors explaining gut retention time (in decimal minutes) when 50% of faeces with experimental foods had been defecated (GRT50%) after 39 feedings of six captive brown bears in Sweden 2010, in relation to diet of berries or carcasses, activity score, midday or midnight feeding, cumulative faeces weight, the interaction term between sex and subadult/adult and with bear ID and experiment as random effects based on the most parsimonious linear mixed model (see Table 2). Variances of random effects are < 0.0 for bear ID, 7,965.2 for experiment, and 35,474.4 for residuals. Markov Chain Monte Carlo (MCMC)-simulated β and its 95% highest posterior density interval (HPD) are given with β and standard errors (SE) based on a t-distribution.
Descriptive estimates for fixed factors used to analyse effects on gut retention time on six captive bears after having a diet comprised of either berries (20 feedings) or carcasses (19 feedings). Bears were either given their experimental foods at midday (N = 22) or at midnight (N = 17).