Behavioral Ecology Advance Access originally published online on July 24, 2008
Behavioral Ecology 2008 19(6):1258-1266; doi:10.1093/beheco/arn079
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Risk effects in elk: sex-specific responses in grazing and browsing due to predation risk from wolves
Department of Ecology, Montana State University, 310 Lewis Hall, Bozeman, MT 59717, USA
Address correspondence to D. Christianson. E-mail:davidc{at}montana.edu.
Received 1 March 2008; revised 16 May 2008; accepted 28 May 2008.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSTUDY AREA AND METHODSRESULTSDiscussionSUPPLEMENTARY MATERIALREFERENCES |
|---|
Risk effects in ungulates are poorly understood but have recently been implicated as an important driver of elk (Cervus elaphus) population dynamics since wolves were reintroduced into the Greater Yellowstone Ecosystem (GYE) of Montana and Wyoming, USA. From December to May in 2004, 2005, and 2006, we recorded the daily presence of wolves on 3 sites occupied by the Upper Gallatin elk population in the northwest corner of the GYE. We estimated the proportion of grasses, conifers, evergreen shrubs, and woody stems in 980 elk fecal samples collected from those 3 sites and tested whether wolf presence affected elk diets. The winter of 2005 was extremely mild allowing us the opportunity to investigate how elk–wolf interactions might change if winter snowpack continues to decline in western North America due to global warming. Snow accumulation consistently favored browsing, and diets during the mild winter were dominated by grass, very similar to the spring diet. In normal winters, adult males grazed less than adult females except when wolves were near because females decreased grazing in response to wolves. Adult males decreased browsing on conifers by half whereas adult females doubled conifer browsing on days when wolves were near. Overall, the sexes had different diets when wolves were absent but showed strong overlap when wolves were present. Diet shifts due to wolves may be causing trophic cascades that have gone unrecognized and probably carry nutritional consequences for wintering elk.
Key words: behaviorally mediate trophic cascade, browsing, diet, elk, grazing, predation, risk effects, sexual segregation, wolves.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREA AND METHODSRESULTSDiscussionSUPPLEMENTARY MATERIALREFERENCES |
|---|
The impact of predators on prey populations has been extensively studied in a wide range of taxa (Begon et al. 1996
). When top-down limitation is thought to be important, predators are commonly assumed to limit prey through direct predation (Sinclair et al. 1998; Connolly and Roughgarden 1999; Hanski et al. 2001). However, a growing body of work shows that predators can also regulate prey through risk effects or reductions in prey survival, growth, and reproduction due to the costs of behavioral or physiological responses to predation (Lima 1998; Werner and Peacor 2003). Risk effects are often manifested through changes in foraging time or behavior (Lima and Dill 1990; Schmitz et al. 1997), but relatively few studies have directly tested the consequent effects on the diet of the prey (but see Beckerman et al. 1997; Banks et al. 1999; Hodges and Sinclair 2003; Eklov and Svanback 2006). Changes in foraging behavior can also affect lower trophic levels (e.g., primary producers) as prey alter intake rates or diet selection patterns (Schmitz et al. 2004). However, it is unclear if such changes in foraging behavior due to antipredator responses can be generalized across systems (Schmitz et al. 2000; Shurin et al. 2002). Because much of our knowledge of risk effects comes from experiments, often with small animals, it is especially unclear whether risk effects can depress the fitness of large terrestrial vertebrates foraging in natural systems (Creel and Christianson 2008).
Foraging behavior in ungulates is often dichotomized between grazing and browsing that are functionally very different foraging strategies, each with unique constraints and nutritional consequences (Ngugi et al. 1995; Clauss et al. 2003). Empirically derived averages of grass and browse in the diet are often a primary characteristic used in herbivore classification (Hofmann 1989). Attempts to uncover the significance of herbivores’ diets have focused strongly on fixed, predefined levels of grazing and browsing, leading to an abundance of theories (Robbins et al. 1995; Owen-Smith 1997; Janis et al. 2000; Mysterud et al. 2001; Perez-Barberia and Gordon 2001; Clauss et al. 2003). Intraspecific variation in the balance of grazing and browsing (and its causes) has received less attention but diet shifts along the grazer–browser continuum may have important consequences for nutrition, competitive interactions, trophic cascades, and diversification. Particularly for intermediate feeders, identifying what factors cause individuals to oscillate between these grazing and browsing may provide novel insight into herbivore evolution.
The recent reintroduction and conservation of wolves (Canis lupus) that primarily prey on elk (Cervus elaphus) in the Greater Yellowstone Ecosystem (GYE) of Montana, Wyoming, and Idaho, USA, has generated new opportunities to examine how large carnivores interact with an intermediate feeder in a natural system (Smith et al. 2003). Risk effects might be quite strong in determining elk population dynamics in the GYE because wolf predation risk in late winter is negatively correlated with fecal progesterone levels and subsequent calf recruitment across several elk populations (Creel et al. 2007). In the Upper Gallatin elk herd, in the northwestern corner of the GYE, the decline in elk population size and calf recruitment following wolf recolonization cannot be directly accounted for by measured rates of wolf predation (Creel and Christianson 2008). It is not known if these apparent risk effects are driven by nutritional consequences of antipredator responses to wolves or some other mechanism (e.g., chronic elevation of glucocorticoid stress hormones). Wolves alter elk movement patterns (Fortin et al. 2005), decrease group size (Creel and Winnie 2005), increase vigilance levels (Liley and Creel 2007; Winnie and Creel 2007), and decrease sensitivity to environmental conditions (Winnie et al. 2006), and these factors seem likely to alter elk energetics or foraging behavior. Most directly, Creel et al. (2005) found that Upper Gallatin elk responded to the presence of wolves at fine spatial (2–4 km) and temporal scales (
1 day) and used forested habitats more frequently on days when wolves were near. This is important because habitat use is a strong driver of diet selection in elk (see recent review by Christianson and Creel 2007), that is, when elk occupy forested habitats, they browse considerably more often. Finally, predation risk from human hunting has been directly shown to increase the proportion of the diet that is browsed when elk shifted to wooded habitats that were apparently safer (Morgantini and Hudson 1985). Combined, these results suggest the testable hypothesis that elk may graze less and browse more in response to wolves. This hypothesis has not been directly tested, and we know of no studies that have identified the nutritional mechanism when risk effects have been identified as a potential driver of population dynamics in a wild vertebrate.
In addition to direct predation and risk effects, the severity of winter is an important component of temperate herbivore ecology, affecting diet selection, survival, and reproduction (Schaller and Ren 1988; Singer et al. 1997; Garrott et al. 2003; Mysterud and Ostbye 2006; Christianson and Creel 2007). Snow depth and density can strongly constrain foraging by burying grasses (Fancy and White 1985; Schaefer and Messier 1996), whereas cold temperatures and locomotion through snow are energetically costly (Fancy and White 1987). However, warming temperatures and declining snowpack in recent decades have led to increasingly mild winters in western North America (Barnett et al. 2005; Mote et al. 2005). Whereas the effects of predation risk and snow condition may be to favor browsing at fine spatial and temporal scales, increasingly mild winters should favor a trend towards grazing and reduce the nutritional constraints on all foraging decisions. We tested the hypothesis that variation in predation risk from wolves and local snow conditions altered the balance of grass and browse in the diets of elk. We also tested the hypothesis that the effects of predation risk and snow on diet composition may be mediated by overall winter severity. The last hypothesis explores the possibility that community interactions may change as the climate continues to warm and winter snowpack in temperate regions continues to decline (Burns et al. 2003; Wilmers and Getz 2005).
|
STUDY AREA AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREA AND METHODSRESULTSDiscussionSUPPLEMENTARY MATERIALREFERENCES |
|---|
The Upper Gallatin elk herd is an annually migratory population [mean ± standard deviation [SD], 1100 ± 260 animals from aerial counts, 2003–2005 Montana Fish Wildlife and Parks, Bozeman, MT, USA] that moves between alpine and subalpine (>2400 m asl) summer habitats in the Gallatin and Madison mountains of Montana and Wyoming to lower elevation (1975–2200 m asl) foothills and valleys along the upper Gallatin river of Montana, in winter. We conducted this study from December to May when elk congregated on the winter range. The elk winter range is composed of large expanses of open sagebrush-steppe (Artemesia spp.) and grasslands (mostly Agropyron and Festuca spp.) with steeper slopes and higher elevations dominated by mature conifer forest (Pinus contorta, Picea engelmanii, Abies lasiocarpa, and Psuedostuga menziesii). Population monitoring since 1928 confirms a consistent pattern of elk distribution across the Upper Gallatin winter range. Most the elk population divides itself across the 4 most open drainages of the Gallatin river (Brazda 1953
; Peek et al. 1967; Peek and Lovaas 1968; Creel et al. 2005; Winnie and Creel 2007). These 4 drainages form 3 distinct sites (2 smaller drainages are directly adjacent) that are separated from one another by rugged and densely forested terrain and the Gallatin River and highway 191, which run south to north through the middle of the study area (for additional descriptions of the study area, see Creel et al. 2005; Winnie and Creel 2007). Recent and historical data confirm similar densities and little movement by elk between sites in winter (Brazda 1953; Peek and Lovaas 1968; Creel et al. 2005).
Elk diet selection
We stratified data collection effort across the 3 sites from December through May in each of 3 winters: 2004, 2005, and 2006. We collected 3–10 fresh (<24 h) fecal samples (
30 ml) from separate piles of elk fecal pellets in each of the 3 sites approximately every 14 days. We used direct observation of elk and snow tracking to maximize the probability that each collection occurred within a single group of elk and each sample came from a single individual (mean of 8.8 samples/collection). We adapted a microhistological technique to prepare microscope slides of each fecal sample to determine its botanical composition using genus- or species-specific epidermal features of plants (Sparks and Malechek 1968). After blending in a household blender for 6 min and soaking in 95% ethanol for 7 days, we washed and bleached (3% sodium hypochlorite for 6 min) fecal and rumen samples over a #200 sieve. We then soaked a 3-ml subsample in lactophenol blue for 7 days before washing it over the sieve. We spread an approximately 1-mm3 subsample over a 25-mm2 area on a 25 x 75 mm glass slide and protected the mount with a 25-mm2 glass cover slip.
We used our slide preparation technique on dried (48 h at 55 °C) and ground (over a 1-mm screen) forage plants collected from the study area to make 59 reference slides for use in identifying epidermal plant fragments on slides made from fecal samples. An independent laboratory specializing in microhistological diet analysis (B. Davitt, Wildlife Habitat Nutrition Laboratory, Washington State University, Pullman, WA, USA) found 27 plant genera or species (11 graminoids, 7 forbs, 6 shrubs, and 3 trees) in a pooled, random sample of elk feces. Reference slides included these 27 forage plants and 32 other common plants and plant parts (stems and leaves) found on the Upper Gallatin elk winter range. To increase the accuracy of assignments (but reducing specificity), we categorized reference plants into 6 vegetative groups based on similarities in epidermal features. These groups were 1) graminoids, 2) forbs, 3) conifer needles, 4) evergreen shrub leaves (primarily sagebrush, Artemesia spp.), 5) willow/aspen stems (Salix spp. and Populus tremuloides), and 6) other woody stems (mostly Ribes spp., Symphoricarpos spp., Rosa acicularis, conifer and evergreen shrub stems). Epidermal fragments from 4 plants were readily identifiable to the generic or species level and could not be easily confused for other plants: Douglas fir needles (P. menziesii), silverberry stems (Eleagnus commutata), Oregon grape leaves (Berberis repens), and phlox leaves (Phlox spp.). A preliminary analysis involving 112 fecal samples found that 99.97% of epidermal fragments were classified to one of these 10 categories (i.e., only 3 of 11 200 epidermal fragments could not be identified).
For microscopic determination of each diet, we identified (at 100x) the first 30 epidermal fragments intersected by an ocular crosshair while scanning random, 25-mm transects on each slide using a mechanical stage. Transects ran parallel to one another and perpendicular to the long axis of the slide with an interval spacing of 1 mm. To avoid observer bias, all slides were labeled with an identification number, and all diets were tabulated on a spreadsheet used only to tally fragments so that the observer was blind with respect to all independent variables that were subsequently paired with each diet in the statistical analyses (see below). Further, all diets were determined by a single observer over a continuous 6-month period in a haphazardly random, nonsequential order. After each diet was tallied, we added Douglas fir fragments to the total count for conifer fragments and added silverberry and willow/aspen stem fragments to the total count of woody stem fragments. We combined counts of graminoid, forb, Oregon grape, and phlox fragments into a single category to describe grazing—the portion of the diet that is primarily consumed while the head is down, usually in open habitats (Christianson and Creel 2007). Forbs, Oregon grape, and phlox usually formed a very small portion of the diet that was positively correlated with graminoid consumption (see Results) so combining these plants with grass into a single category had little effect on results. For simplicity, we refer to this category as grazing or grass. These procedures yielded 4 broad categories to describe the diet of elk: percent grazing, percent conifers, percent evergreen shrubs, and percent woody stems. These categories are well suited to detect potential changes in elk diets due to shifts in habitat selection or shifts in microsite foraging patterns.
Wolf predation risk
Wolves recolonized the upper Gallatin in 1996 following a wolf reintroduction program in neighboring Yellowstone National Park, WY, USA in 1995–1996 (Fritts et al. 1997). From 2004–2006, wolves persisted across the study area (2–3 packs, 8–17 wolves) and regularly moved between drainages or off the study area, creating temporal and spatial variation in predation risk for elk. We recorded whether wolves were detected or not each day that any data were collected within a site. Radio collars were maintained on 0–4 wolves throughout the study period as part of population monitoring effort (US Fish and Wildlife Service et al. 2005; Sime et al. 2007). Wolf sightings, radiotelemetry, howls, or fresh (<24 h) tracks, kills, or scats confirmed if wolves were present within a site on any given day. We scored wolves as absent from a site if they were not detected by any of these means. Although we undoubtedly failed to detect wolves on some occasions, this method is conservative in that detection failures would tend to underestimate the effects of wolves on the diet. Previously, this method has detected effects of local wolf presence on elk behavior, grouping, habitat selection, and sensitivity to environmental conditions.
In describing effects of predation risk on elk foraging, we recognized that fresh fecal samples contain information on the foraging decisions made by elk over the previous 2–3 days (Jiang and Hudson 1996). To address the time lag between plant consumption and fecal collection, we also considered whether wolves were detected in a site over the 2 days preceding the day of fecal collection. That is, we dichotomized elk diets as being consumed with "wolves present" or "wolves absent" depending on whether or not wolves were detected within the site on the day of, the day preceding, or 2 days preceding fecal collection. If wolves were never detected in those 3 days, we considered wolves to be absent (note that wolf locations are strongly autocorrelated at this spatiotemporal scale, see Bergman et al. 2006). We excluded samples that were collected when no determination of wolf presence or absence had been made.
Snow
Snow can increase the vulnerability of prey to predators by restricting movement (Huggard 1993; Bergman et al. 2006). Snow depth and density can also affect diet selection in herbivores by restricting access to herbaceous vegetation, especially in those intermediate feeders that prefer not only grazing but also browse (Skogland 1978; Telfer and Kelsall 1984; Jenkins and Wright 1987; Adamczewski et al. 1988; Christianson and Creel 2007). We measured snow depth and compaction by dropping a 3-kg steel ball from 30 cm and recording its penetration and total snow depth to the nearest 1 cm at 3 fixed points in each site every 14 days (Winnie et al. 2006). We used penetration and depth to scale snow compaction from 0 (soft, powdered snow) to 1 (impermeable crusts) with the formula, compaction = 1 – (cm of penetration/cm of total snow depth). For each site and 14-day period, we joined the minimum snow depth and maximum compaction with the formula, snow index = snow depthmin + snow depthmin x compactionmax. We used the minimum snow depth (of the 3 collected within each site and 14-day period) because elk can forage selectively and will preferentially select areas free of snow due to wind or snow melt (Houston 1982) and mean or maximum snow depth could fail to capture this ecological circumstance. For each site, we matched each fecal collection with the snow index measurement nearest in time (Figure 1).
|
Sex
Elk are sexually dimorphic, polygynous ungulates, and adults assort into herds with highly biased sex ratios (often single-sex herds) in the Upper Gallatin in winter (Peek and Lovaas 1968
; Creel and Winnie 2005; Creel et al. 2005; Winnie and Creel 2007). This sexual segregation was used to classify fecal samples according to sex based on the sex bias of groups or sites and fecal pellet size (Vales and Peek 1996; Bleich et al. 1997). Frequently (58.57% of all fecal samples), the composition of a group (number of cows, calves, yearling males, and adult males) was directly observed immediately prior to fecal collection. In addition, we walked fixed transects every 14 days in each drainage and used systematic scans of the entire viewshed with binoculars and spotting scopes to locate and classify every elk group. While regularly collecting other data on elk foraging behavior, wolf predation, and environmental conditions throughout the study area, we also scanned for elk, counting, and classifying every group detected. If the group or drainage composition was directly observed before fecal collection, we classified samples according to sex based on the proportion of adult males (‘bull’ if >85% or ‘cow/calf’ if <15%). If the group or recent drainage compositions were of mixed sex (85% 
% male 15%) at the time of fecal collection, we used fecal pellet size to distinguish between the sex of individuals (Maccracken and Vanballenberghe 1987; Khan and Goyal 1993). Sex-specific pellet sizes were determined from probability density functions of the pellet size for each sex/age class. We developed probability density functions of fecal pellet size for calves, adult females, and adult males based on fecal pellets collected from carcasses found opportunistically on the study site (see supplementary material—Appendix B). Sexual segregation was pronounced, so most samples (88.0%) could be classified using the group or drainage sex bias: of 13104 elk in 1138 groups classified over the 3 winters, 92% of all elk were in groups that were either >90% or <10% adult male.
Data analysis
Broad inter- and intra-annual variation in environmental conditions can interact with elk behavioral responses (Singer and Norland 1994; Singer and Harter 1996; Vales and Peek 1996; Gude et al. 2006). Further, climate change is predicted to affect herbivores through changes in winter energy demands and forage plant phenology (Chan et al. 2005; Pettorelli et al. 2005; Mysterud and Ostbye 2006). The winter of 2005 was arguably the mildest in nearly 100 years of records for the Rocky Mountains (Saunders and Maxwell 2005). The winter of 2005 was one of the lowest years for April 1 snowpack in 47 years of record keeping in the Upper Gallatin, an anomaly related to the long-term trend toward milder winters with less snow on this and other elk winter ranges (Appendix A and see Wilmers and Getz 2005). Because the winter of 2005 was exceptionally mild (Figure 1), we separated samples collected in this "mild winter" from the "normal winters" of 2004 and 2006 and conducted a separate analysis on each winter type. With this approach, we examined how diets and elk–wolf interactions are affected by the background level of winter severity that provided insight on how these relationships may change due to increasing atmospheric temperatures and declining winter snowpack (Barnett et al. 2005).
Although our primary focus was to test for dietary responses to wolves in normal winters when the nutritional consequences of constraints on foraging may be most significant, we found that an analysis of spring diets provided an important contrast that facilitated interpretation of elk foraging ecology. Elk life history and population dynamics appear strongly linked to the period when snow cover no longer constrains grazing and new plant growth provides a highly nutritious alternative to the dormant culms and leaves that sustain elk through winter (Singer et al. 1997; Vore and Schmidt 2001; Taper and Gogan 2002). However, defining the onset of spring in terms relevant to herbivory has often proven difficult. Fixed Julian dates, such as the vernal equinox, can be arbitrary and do not allow for substantial latitudinal, altitudinal, or annual variation in the rate of snowmelt or the onset and rate of spring growth (Pettorelli et al. 2005). Additionally, regardless of when spring may "arrive," herbivores vary considerably in their degree of selectivity for new spring growth so that the demarcation between winter and spring nutrition may be quite different across species in similar seasonal environments. To separate spring samples from winter samples, we used the concentration of chlorophyll in elk fecal samples as measured by spectrophotometry of fecal extracts (Christianson 2008). Specifically, we dichotomized diets based on their collection date by using the day after which fecal chlorophyll concentrations showed persistently elevated levels, indicative of increased intake of photosynthetically active plant tissue. To determine this date, we used the breakpoint estimated by a piecewise regression of fecal chlorophyll onto Julian day in each winter. Mean chlorophyll concentration between winter and spring samples differed by a factor of 12 so this demarcation date was unambiguous and clearly defined in each year. The date [±95% confidence interval (CI)] of demarcation between winter and spring differed significantly across years: March 28 (±2.3 days) in 2004, April 1 (±2.0 days) in 2005, and April 21 (±1.3 days) in 2006, and unlike traditional descriptions of the onset of spring, this measure described the date at which elk begin to consume green biomass in quantity (Christianson 2008). In this manner, we identified the timing of spring green-up as it actually affected elk diets.
We determined the diet composition for 980 fecal samples. A visual assessment of the normalized response variables (percent grazing, percent conifers, percent evergreen shrub, or percent woody stems) confirmed nonnormality due to biases in the frequency of diets toward 0 and 100%. Therefore, we transformed the response variables using the logit transformation on adjusted proportions (Fox 1997). We used analysis of covariance (ANCOVA) within each winter type (normal and mild) to explain variation in each diet component as a factor of wolf presence and elk sex while controlling for the snow index, a continuous covariate. We used analysis of variance (ANOVA) to explain variation in spring diets considering the effect of wolves and elk sex. Wolf predation risk within sites varied considerably (Figure 2) providing high power to test for the effects of wolf predation risk on elk diets with ANCOVA (winters) and ANOVA (spring). We determined 401 diets for samples collected at times when wolves had not been detected and 420 diets collected at times when wolves had been detected. We could not confirm wolf presence or absence on the day of, or the 2 days preceding fecal collection for 159 samples, and these samples were excluded from ANCOVA's and ANOVA's. We collected samples from 255 bulls and 708 cow/calves or a ratio of 2.77 cow/calf samples:bull samples (sex could not be determined for 17 samples). These sample sizes closely mirrored the sex ratio of the population from composition counts (2.68 cow/calf:bull), and we interpret our results as extrapolative to the population level. We standardized all variables before analysis to allow direct comparison of their effect sizes. Because logit-transformed means cannot be directly interpreted and confidence intervals from logit-transformed data are asymmetric, we report back-transformed 95% CI from ANCOVA's (with snow controlled at its mean) or ANOVA in the text for simplicity and to facilitate comparisons and interpretations of biological significance.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREA AND METHODSRESULTSDiscussionSUPPLEMENTARY MATERIALREFERENCES |
|---|
General patterns in elk diet selection
Within this single population, elk showed a broad scope of variation in their use of grazing and browsing strategies: diet ranges including 0–100% grazing, 0–90% conifers, 1–53% evergreen shrubs, and 0–100% woody stems. As is typical of elk, graminoids formed the bulk of the winter diet (83.3% of samples contained >50% graminoids). The least grazing occurring in normal winters and the most grazing in the extremely mild winter of 2005 and in spring (Table 1): this effect is also typical of elk. Forbs (mean ± SD, 4.0 ± 7.1%), Oregon grape (2.1 ± 5.0%), and phlox (0.5 ± 1.6%) always made small contributions to the diet (Table 1) and were positively correlated with graminoids in the diet (r = 0.068, P = 0.034) providing justification for combining these variables into a slightly broader diet component, grazing. In spring, grazing heavily dominated the diet in 2004, 2005, and 2006 (91.4 ± 9.7%, n = 108; 96.3 ± 4.0%, n = 101; 92.3 ± 16.5%, n = 44; respectively) and spring sample sizes were smaller than in winter so we pooled spring samples across years.
|
Annual differences in snow conditions followed regional trends, with snow accumulation in the mild winter of 2005 (snow index ± SD, 18.5 ± 16.8) only half of that in the normal winters of 2004 and 2006 (Figure 1, 38.3 ± 23.0, t277,450 = 12.41, P < 0.001). Elk diets were strongly affected by snow in several ways. Increasing snow index reduced grazing in both normal and mild winters (F1,366 = 82.52, P < 0.001 and F1,223 = 41.19, P < 0.001). In parallel, grazing in the mild winter formed a higher proportion of the diet than in normal winters (Table 1, t277,450 = 13.62, P < 0.001) and grazing in spring was higher than grazing in all winters (Table 1, t253,727 = 18.06, P < 0.001). Snow had the strongest effect on the diet in most models (Table 2).
|
Elk diet selection in normal winters
In normal winters, ANCOVA showed that cow/calf groups decreased grazing in response to wolves (back transformed 95% CI, see Study area and methods: 75.6–83.0%, wolves absent vs. 61.6–69.4%, wolves present) whereas bulls showed no response in grazing (56.3–69.7%, wolves absent vs. 60.3–72.1%, wolves present), so the sex x wolf interaction was strong (F1,366 = 11.67, P < 0.001). Bull elk, however, decreased conifer use by half on days when wolves shared the same site (18.7–30.4%, wolves absent vs. 8.4–14.7%, wolves present); in contrast cow/calf groups nearly doubled conifer use (4.4–6.6% wolves absent vs. 7.8–11.4% wolves present). Consequently, the main effect of wolves on conifer use in normal winters not important (F1,366 = 1.22, P = 0.271) but the sex x wolf interaction was very strong (F1,366 = 30.34, P < 0.001). These opposing responses to wolves would have been missed if the sexes had not been separately examined (Figure 3). The sexes also differed in their consumption of evergreen shrubs in normal winters with cow/calf groups consuming more than bulls (Figure 3, sex effect F1,366 = 13.76, P < 0.001), but both sexes tended to increase consumption of sagebrush when wolves were present (wolf effect, F1,366 = 27.13, P < 0.001; cow/calf groups: 3.9–5.4%, wolves absent vs. 7.6–10.9%, wolves present; bull groups 2.9–4.2%, wolves absent vs. 4.1–6.6%, wolves present). Thus, the sex x wolf interaction did not significantly contribute to evergreen browse consumption (F1,366 = 0.73, P = 0.394). This wolf effect on evergreen shrub use was stronger than the effect of snow in normal winters (Table 2). Wolves also positively (but less strongly than snow) affected consumption of woody stems (F1,366 = 4.68, P = 0.031), but sex differences were not apparent (sex effect: F1,366 = 0.13, P = 0.715; sex x wolf effect: F1,366 = 1.06, P = 0.303).
|
Elk diet selection in a mild winter and spring
Broad diet composition, sex differences, and responses to wolves were very similar between the extremely mild winter of 2005 and the spring of all 3 years, and diets under both circumstances were quite different from normal winters (Figure 3). Diets were more strongly biased by sex in the mild winter and spring, as cows grazed considerably more than bulls (sex effect, mild winter only, F1,223 = 80.02, P < 0.001; spring, F1,223 = 29.15, P < 0.001). Cows did not decrease grazing in response to wolves in the mild winter, as they had in the normal winters (89.1–92.2%, wolves absent vs. 91.7–94.3%, wolves present). Somewhat surprisingly, bulls increased grazing on days when wolves shared the same site (60.7–75.7%, wolves absent vs. 80.7–87.3%, wolves present), whereas in normal winters the response was not important (Figure 3). Consequently, there was no difference in the sign on the sex x wolf coefficient between normal and mild winters (Table 2; mild winters, F1,223 = 5.26, P < 0.023), even though the detailed patterns that produced this interaction were quite different (Figure 3). This sex-dependent response in grazing was similar to that observed in spring (Figure 3, cow/calf groups 95.6–96.6%, wolves absent vs. 95.8–97.0% wolves present; bulls 74.0–92.7%, wolves absent vs. 89.3–95.0% wolves present) though the sex x wolf interaction was weaker in spring (F1,253 = 1.67, P = 0.198). The main effect of wolves on grazing was marginally significant in spring (F1, 253 = 3.49, P = 0.063), and cow/calf groups generally failed to show any responses to wolves in consumption of individual browse components in the mild winter or spring (Figure 3). In response to wolf presence, bulls tended to evenly decrease consumption of conifers and evergreen shrubs in the mild winter of 2005 (conifers: 10.1–22.2% wolves absent vs. 5.9–10.0% wolves present and evergreen shrubs: 3.9–8.2% wolves absent vs. 3.4–4.8% wolves present) and in spring (conifers: 4.8–12.3% wolves absent vs. 3.1–4.3% wolves present and evergreen shrubs: 1.9–3.9% wolves absent vs. 0.8–2.7% wolves present).
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREA AND METHODSRESULTSDiscussionSUPPLEMENTARY MATERIALREFERENCES |
|---|
Perhaps the simplest generalization of wolf effects on elk diet is that wolf presence led the sexes to consume similar diets. This is important because the sexes were strongly segregated (dietarily) in the absence of wolves and remained spatially segregated in the presence of wolves. Each sex's diet responded to wolves differently and seasonal conditions determined that sex responded (Figure 3), but regardless of which sex responded, strongly overlapping diets were the ultimate result of foraging under the risk of predation. Previous research (Winnie et al. 2006
) showed that the elk gender and extrinsic factors became weaker predictors of elk habitat selection when wolves were present in the same drainage. Our results show that this behavioral response to risk leads to a more uniform diet across the elk population. This adds further complexity to our current understanding of sexual segregation as we confirmed that factors related to predation risk cannot be disentangled from factors related to forage acquisition (Mcnamara and Houston 1987) because both were uniquely related in each sex. Perhaps predation risk reinforces spatial segregation because the presence of wolves would exacerbate asymmetric competition for forage between sexes, a new hypothesis presented by our data. However, because the greatest dietary differences between sexes were seen when constraints on foraging were weakest (Figure 3, no wolves in the mild winter or spring) then physiological or allometric differences may ultimately be driving segregation (Loe et al. 2006). Mean diet composition, responses to wolves, and sex differences were strongly mediated by seasonal conditions, most likely through effects of snow on the availability of forage grasses. Our results confirm that snow conditions are extremely important in determining how herbivores use vegetation. Not surprisingly, snow had stronger (negative) effects on grazing in normal winters than in the mild winter, and snow always positively affected the consumption of all types of browse. In the very mild winter, elk showed very high levels of grazing, similar to spring, although spring grazing may also be strongly driven by changes in preference (rather than availability) as nutritious, green grass shoots emerge. The differences in responses to wolves and snow between winters strongly suggest that the effects of climate change on trophic interactions may not be simple linear extensions of trends found in current or historical winters. Decreasing snowpack and warmer temperatures will not only change broad diet patterns in herbivores but may also alter how herbivores interact with conspecifics and respond to predators.
Consequences for elk–plant interactions were also suggested by our results. Others have noted that hardwood species in the GYE have experienced enhanced growth coinciding with wolf reintroduction, and it has been suggested that this pattern reflects a behaviorally-mediated trophic cascade caused by a reduction in browsing behavior by elk in response to wolves (Ripple and Beschta 2004). However, our study is the first to examine diet selection by elk in response to predation risk from wolves. Our results suggest the hypothesis that behaviorally-mediated trophic cascades may be occurring at finer temporal scales in important forage plants such as grasses, conifers, and evergreen shrubs. For example, if adult females in the Upper Gallatin elk population (n 720) reduce grazing from 80 to 66% of the diet on days when wolves are present, and each elk consumes 4.5 kg of primary production/day, and wolves are present for 50% of 120 days in winter, then, each winter, approximately 27 fewer metric tons of herbaceous vegetation are being grazed in foraging areas occupied by adult females since wolf reintroduction. We caution that this would only be true if intake rates do not change drastically in the presence of wolves. We currently know little about how the natural variation in predation risk may be shaping important plant communities in areas recolonized by top predators but given the dominant role of elk in this ecosystem, community-wide responses seem likely (Mclaren and Peterson 1994; Croll et al. 2005).
Finally, the diets of elk responded to fine-scale temporal and spatial variation in predation risk from wolves as part of a broad suite of behavioral responses based on dynamic assessments of risk (Creel and Winnie 2005; Creel et al. 2005; Winnie et al. 2006; Liley and Creel 2007; Winnie and Creel 2007). Adult females increased browsing of conifers in normal winters in response to wolves, matching predictions based on habitat shifts toward forested areas when wolves are near (Creel et al. 2005). Despite suffering disproportionately higher rates of wolf predation than adult females (Winnie and Creel 2007, Christianson 2008), adult male diets responded only weakly to wolves, as has been found previously in other antipredator responses in bulls (Liley and Creel 2007, Winnie and Creel 2007). Winnie and Creel (2007) found that adult males were in worse condition than adult females in winter and suggested that males respond weakly to wolves because behavioral responses might carry nutritional costs that males are less able to meet. Supporting this hypothesis, we found that when wolves were present, bulls showed little change in their use of grass, the most important food item, except in the mild winter and spring (Figure 3) when nutritional constraints were likely much weaker. This logic suggests that cows and calves probably pay fitness costs for the magnitude of their antipredator response in normal winters; indeed the forages cows and calves selected vary considerably in their physicochemical structure and effects on nutrient assimilation (Hoffmann 1989; Ngugi et al. 1995, Robbins et al. 1995; Owen-Smith 1997; Clauss et al. 2003). A nutritional cost is also consistent with the observation that predation risk is strongly correlated with altered reproductive physiology and reduced calf production in elk following wolf recolonization (Creel et al. 2007; Creel and Christianson 2008). In summary, these results confirm one more segment of the pathway that must exist if predation risk can alter prey population dynamics through behavioral mechanisms that carry costs. To date, this pathway has been described primarily by experiments with small animals (Lima 1998; Werner and Peacor 2003). Future research should quantify the potential effect of predation risk on nutrient balance and physical condition of prey in natural systems (Banks et al. 1999). The potential importance of risk effects is well established but we still know little about the relative magnitudes of risk effects and direct predation in natural systems.
|
SUPPLEMENTARY MATERIAL |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREA AND METHODSRESULTSDiscussionSUPPLEMENTARY MATERIALREFERENCES |
|---|
Supplementary material can be found at http://www.beheco.oxfordjournals.org/.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREA AND METHODSRESULTSDiscussionSUPPLEMENTARY MATERIALREFERENCES |
|---|
Adamczewski JZ, Gates CC, Soutar BM, Hudson RJ. Limiting effects of snow on seasonal habitat use and diets of caribou (Rangifer tarandus groenlandicus) on Coats Island, Northwest Territories, Canada. Can J Zool (1988) 66:1986–1996.
Banks PB, Hume ID, Crowe O. Behavioral, morphological, and dietary response of rabbits to predation risk from foxes. Oikos (1999) 85:247–256.[CrossRef][Web of Science]
Barnett TP, Adam JC, Lettenmaier DP. Potential impacts of a warming climate on water availability in snow-dominated regions. Nature (2005) 438:303–309.[CrossRef][Web of Science][Medline]
Beckerman AP, Uriarte M, Schmitz OJ. Experimental evidence for a behavior-mediated trophic cascade in a terrestrial food chain. Proc Natl Acad Sci USA. (1997) 94:10735–10738.
Begon M, Harper JL, Townsend CR. Ecology: individuals, populations and communities. (1996) Vol. 3. Oxford: Blackwell Science.
Bergman EJ, Garrott RA, Creel S, Borkowski JJ, Jaffe R, Watson EGR. Assessment of prey vulnerability through analysis of wolf movements and kill sites. Ecol Appl (2006) 16:273–284.[CrossRef]
Bleich VC, Bowyer R, Wehausen JD. Sexual segregation in mountain sheep: resources or predation? Wildl Monogr (1997) 134:1–50.
Brazda AR. Elk migration patterns, and some of the factors affecting movements in the Gallatin River drainage, Montana. J Wildl Manage (1953) 17:9–23.[CrossRef]
Burns CE, Johnston KM, Schmitz OJ. Global climate change and mammalian species diversity in U.S. national parks. Proc Nat Acad Sci USA. (2003) 100:11474–11477.
Chan KS, Mysterud A, Oritsland NA, Severinsen T, Stenseth NC. Continuous and discrete extreme climatic events affecting the dynamics of a high-arctic reindeer population. Oecologia (2005) 145:556–563.[CrossRef][Web of Science][Medline]
Christianson D. Risk effects in elk: behavioral and nutritional responses to wolves and environmental conditions [dissertation] (2008) Bozeman: Montana State University.
Christianson DA, Creel S. A review of environmental factors affecting elk winter diets. J Wildl Manage (2007) 71:164–176.[CrossRef]
Clauss M, Lechner-Doll M, Streich WJ. Ruminant diversification as an adaptation to the physicomechanical characteristics of forage. A reevaluation of an old debate and a new hypothesis. Oikos (2003) 102:253–262.[CrossRef][Web of Science]
Connolly SR, Roughgarden J. Theory of marine communities: competition, predation, and recruitment-dependent interaction strength. Ecol Monogr (1999) 69:277–296.[Web of Science]
Creel S, Christianson D, Liley S, Winnie JA. Predation risk affects reproductive physiology and demography of elk. Science (2007) 315:960.
Creel S, Winnie J, Maxwell B, Hamlin K, Creel M. Elk alter habitat selection as an antipredator response to wolves. Ecology (2005) 86:3387–3397.[CrossRef][Web of Science]
Creel S, Winnie JA. Responses of elk herd size to fine-scale spatial and temporal variation in the risk of predation by wolves. Anim Behav (2005) 69:1181–1189.[CrossRef][Web of Science]
Creel S, Christianson D. Relationships between direct predation and risk effects. Trends Ecol Evol. (2008) 23:194–201.[CrossRef][Medline]
Croll DA, Maron JL, Estes JA, Danner EM, Byrd GV. Introduced predators transform subarctic islands from grassland to tundra. Science (2005) 307:1959–1961.
Eklov P, Svanback R. Predation risk influences adaptive morphological variation in fish populations. Am Nat (2006) 167:440–452.[CrossRef][Web of Science][Medline]
Fancy SG, White RG. Energy expenditures by caribou while cratering in snow. J Wildl Manage (1985) 49:987–993.[CrossRef]
Fancy SG, White RG. Energy expenditures for locomotion by barren-ground caribou. Can J Zool (1987) 65:122–128.
Fortin D, Beyer HL, Boyce MS, Smith DW, Duchesne T, Mao JS. Wolves influence elk movements: behavior shapes a trophic cascade in Yellowstone National Park. Ecology (2005) 86:1320–1330.[CrossRef][Web of Science]
Fox J. Applied regression analysis, linear models, and related methods (1997) Thousand Oaks (CA): Sage Publications.
Fritts SH, Bangs EE, Fontaine JA, Johnson MR, Phillips MK, Koch ED, Gunson JR. Planning and implementing a reintroduction of wolves to Yellowstone National Park and central Idaho. Restoration Ecol (1997) 5:7–27.[CrossRef]
Garrott RA, Eberhardt LL, White PJ, Rotella J. Climate induced variation in vital rates of an unharvested large-herbivore population. Can J Zool (2003) 81:33–45.[CrossRef]
Gude JA, Garrott RA, Borkowski JJ, King F. Prey risk allocation in a grazing ecosystem. Ecol Appl (2006) 16:285–298.[CrossRef]
Hanski I, Henttonen H, Korpimaki E, Oksanen L, Turchin P. Small-rodent dynamics and predation. Ecology (2001) 82:1505–1520.[CrossRef][Web of Science]
Hodges KE, Sinclair ARE. Does predation risk cause snowshoe hares to modify their diets? Can J Zool (2003) 81:1973–1985.[CrossRef]
Hofmann RR. Evolutionary steps of ecophysiological adaptation and diversification of ruminants: a comparative view of their digestive system. Oecologia (1989) 78:443–457.[CrossRef][Web of Science]
Houston DB. The Northern Yellowstone elk (1982) New York: Macmillian.
Huggard DJ. Effect of snow depth on predation and scavenging by gray wolves. J Wildl Manage (1993) 57:382–388.[CrossRef]
Janis CM, Damuth J, Theodor JM. Miocene ungulates and terrestrial primary productivity: where have all the browsers gone? Proc Natl Acad Sci USA. (2000) 97:7899–7904.
Jenkins KJ, Wright RG. Dietary niche relationships among cervids relative to winter snowpack in northwestern Montana, USA. Can J Zool (1987) 65:1397–1401.
Jiang Z, Hudson RJ. Digestive response of wapiti (Cervus elaphus Canadensis) to seasonal forages. Acta Theriol (1996) 41:415–423.
Khan I, Goyal SP. Fecal pellet measurements of manipur brow-antlered deer with respect to sex and age classes. Zoo Biol (1993) 12:561–565.[CrossRef]
Liley S, Creel S. What best explains vigilance in elk: characteristics of prey, predators, or the environment? Behav Ecol (2007) 19:245–254.[CrossRef][Web of Science]
Lima SL. Nonlethal effects in the ecology of predator-prey interactions—what are the ecological effects of anti-predator decision-making? Bioscience (1998) 48:25–34.[CrossRef][Web of Science]
Lima SL, Dill LM. Behavioral decisions made under the risk of predation—a review and prospectus. Can J Zool (1990) 68:619–640.[CrossRef]
Loe LE, Irvine RJ, Bonenfant C, Stien A, Langvatn R, Albon SD, Mysterud A, Stenseth NC. Testing five hypotheses of sexual segregation in an arctic ungulate. J Anim Ecol (2006) 75:485–496.[CrossRef][Medline]
Maccracken JG, Vanballenberghe V. Age related and sex related differences in fecal pellet dimensions of moose. J Wildl Manage (1987) 51:360–364.[CrossRef]
Mclaren BE, Peterson RO. Wolves, moose, and tree rings on Isle Royale. Science (1994) 266:1555–1558.
Mcnamara JM, Houston AI. Starvation and predation as factors limiting population size. Ecology (1987) 68:1515–1519.[CrossRef][Web of Science]
Morgantini LE, Hudson RJ. Changes in diets of wapiti during a hunting season. J Range Manage (1985) 38:77–79.[CrossRef]
Mote PW, Hamlet AF, Clark MP, Lettenmaier DP. Declining mountain snowpack in western North America. Bull Am Meteorol Soc (2005) 86:39–49.[CrossRef]
Mysterud A, Ostbye E. Effect of climate and density on individual and population growth of roe deer Capreolus capreolus at northern latitudes: the Lier valley, Norway. Wildl Biol (2006) 12:321–329.[CrossRef]
Mysterud A, Perez-Barberia FJ, Gordon IJ. The effect of season, sex and feeding style on home range area versus body mass scaling in temperate ruminants. Oecologia (2001) 127:30–39.[CrossRef][Web of Science]
Ngugi RK, Hinds FC, Powell J. Mountain big sagebrush browse decreases dry-matter intake, digestibility, and nutritive quality of sheep diets. J Range Manage (1995) 48:487–492.[CrossRef]
Owen-Smith N. Distinctive features of the nutritional ecology of browsing versus grazing ruminants. Z Saeugetierkd (1997) 62:176–191.
Peek JM, Lovaas AL. Differential distribution of elk by sex and age on the Gallatin winter range, Montana. J Wildl Manage (1968) 32:553–557.[CrossRef]
Peek JM, Lovaas AL, Rouse RA. Population changes within the Gallatin elk herd 1932–65. J Wildl Manage (1967) 31:304–316.[CrossRef]
Perez-Barberia FJ, Gordon IJ. Relationships between oral morphology and feeding style in the Ungulata: a phylogenetically controlled evaluation. Proc R Soc Lond B Biol Sci. (2001) 268:1023–1032.[Medline]
Pettorelli N, Vik JO, Mysterud A, Gaillard JM, Tucker CJ, Stenseth NC. Using the satellite-derived NDVI to assess ecological responses to environmental change. Trends Ecol Evol. (2005) 20:503–510.[CrossRef][Medline]
Ripple WJ, Beschta RL. Wolves, elk, willows, and trophic cascades in the upper Gallatin Range of Southwestern Montana, USA. For Ecol Manage (2004) 200:161–181.[CrossRef]
Robbins CT, Spalinger DE, Vanhoven W. Adaptation of ruminants to browse and grass diets—are anatomical-based browser-grazer interpretations valid. Oecologia (1995) 103:208–213.[CrossRef][Web of Science]
Saunders S, Maxwell M. Less snow, less water: climate disruption in the west. (2005) Louisville (CO): The Rocky Mountain Climate Organization.
Schaefer JA, Messier F. Winter activity of muskoxen in relation to foraging conditions. Ecoscience (1996) 3:147–153.[Web of Science]
Schaller GB, Ren JR. Effects of a snowstorm on Tibetan antelope. J Mammal (1988) 69:631–634.[CrossRef][Web of Science]
Schmitz OJ, Beckerman AP, Obrien KM. Behaviorally mediated trophic cascades: effects of predation risk on food web interactions. Ecology (1997) 78:1388–1399.[CrossRef][Web of Science]
Schmitz OJ, Hambaeck PA, Beckerman AP. Trophic cascades in terrestrial systems: a review of the effects of carnivore removals on plants. Am Nat (2000) 155:141–153.[CrossRef][Medline]
Schmitz OJ, Krivan V, Ovadia O. Trophic cascades: the primacy of trait-mediated indirect interactions. Ecol Lett (2004) 7:153–163.[CrossRef][Web of Science]
Shurin JB, Borer ET, Seabloom EW, Anderson K, Blanchette CA, Broitman B, Cooper SD, Halpern BS. A cross-ecosystem comparison of the strength of trophic cascades. Ecol Lett (2002) 5:785–791.[CrossRef][Web of Science]
Sime C, Asher V, Bradley L, Laudon K, Ross M, Trapp J, Atkinson M, Handegard L, Steuber J. Montana gray wolf conservation and management 2006 annual report (2007) Helena (MT): Montana Fish, Wildlife, and Parks.
Sinclair ARE, Pech RP, Dickman CR, Hik D, Mahon P, Newsome AE. Predicting effects of predation on conservation of endangered prey. Conserv Biol (1998) 12:564–575.[CrossRef]
Singer FJ, Harter MK. Comparative effects of elk herbivory and 1988 fires on northern Yellowstone National Park grasslands. Ecol Appl (1996) 6:185–199.[CrossRef]
Singer FJ, Harting A, Symonds KK, Coughenour MB. Density dependence, compensation, and environmental effects on elk calf mortality in Yellowstone National Park. J Wildl Manage (1997) 61:12–25.[CrossRef]
Singer FJ, Norland JE. Niche relationships within a guild of ungulate species in Yellowstone National Park, Wyoming, following release from artificial controls. Can J Zool (1994) 72:1383–1394.[CrossRef]
Skogland T. Characteristics of snow cover and its relationship to wild mountain reindeer (Rangifer tarandus tarandus) feeding strategies. Arc Alp Res. (1978) 10:569–579.[CrossRef]
Smith DW, Peterson RO, Houston DB. Yellowstone after wolves. Bioscience (2003) 53:330–340.[CrossRef][Web of Science]
Sparks DR, Malechek JC. Estimating percentage dry weight in diets using a microscopic technique. J Range Manage (1968) 21:264–265.[CrossRef]
Taper ML, Gogan PJP. The Northern Yellowstone elk: density dependence and climatic conditions. J Wildl Manage (2002) 66:106–122.[CrossRef]
Telfer ES, Kelsall JP. Adaptation of some large North American mammals for survival in snow. Ecology (1984) 65:1828–1834.[CrossRef][Web of Science]
U.S. Fish and Wildlife Service, Nez Perce Tribe, National Park Service, Montana Fish Wildlife and Parks, Idaho Fish and Game. Rocky Mountain wolf recovery 2004 annual report (2005) Helena (MT): US Fish and Wildlife Service.
Vales DJ, Peek JM. Responses of elk to the 1988 Yellowstone fires and drought. In: The Ecological implications of fire in the Greater Yellowstone: second biennial conference on science in the Greater Yellowstone ecosystem—Greenlee JM, ed. (1996) Fairfield (WA): International Association of Wildland Fire.
Vore JM, Schmidt EM. Movements of female elk during calving season in northwest Montana. Wildl Soc Bull. (2001) 29:720–725.
Werner EE, Peacor SD. A review of trait-mediated indirect interactions in ecological communities. Ecology (2003) 84:1083–1100.[CrossRef][Web of Science]
Wilmers CC, Getz WM. Gray wolves as climate change buffers in Yellowstone. PLoS Biol (2005) 3:571–576.[Web of Science]
Winnie J, Christianson D, Creel S, Maxwell B. Elk decision-making rules are simplified in the presence of wolves. Behav Ecol Sociobiol. (2006) 61:277–289.[CrossRef][Web of Science]
Winnie J, Creel S. Sex-specific behavioural responses of elk to spatial and temporal variation in the threat of wolf predation. Anim Behav (2007) 73:215–725.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Creel, J. A. Winnie Jr., and D. Christianson Glucocorticoid stress hormones and the effect of predation risk on elk reproduction PNAS, July 28, 2009; 106(30): 12388 - 12393. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||