A NOVEL METHOD OF PERFORMING MOOSE BROWSE SURVEYS Rachel L.W. Portinga1,2,3 and Ron A. Moen1,2 1Biology Department, University of Minnesota Duluth, 1049 University Drive, Duluth, MN 55812; 2University of Minnesota Duluth Natural Resources Research Institute, 5013 Miller Trunk Hwy, Duluth, MN, 55811 ABSTRACT: We measured browse availability and use along foraging paths of GPS radio-collared moose (Alces alces) in northeastern Minnesota to estimate diet composition and browse species pref- erence. On foraging paths during summer and winter we counted twigs via traditional methods for comparison with a novel method that attempted to better simulate moose foraging behavior. Twigs were collected and used to develop diameter at point of browsing – biomass regressions for each browse species. These regressions, different under open and closed canopy, were used to estimate bio- mass consumption on foraging paths and to compare 4 approaches. The average diets were similar to previously measured regional diets, and importantly, our data identified variance among individual seasonal diets. Our field method allowed us to better quantify and compare diet composition and browse selection of individual free-ranging moose directly on foraging paths. ALCES VOL. 51: 107–122 (2015) Key words: bite size, browse availability method, browse selection, diameter-biomass regressions, diet composition, Minnesota Large herbivores like moose (Alces alces) view their food resources at the land- scape, patch, and feeding station levels (Senft et al. 1987). At the landscape level moose choose which patches to visit based on the spatial distribution of browse density and forage availability within each patch. At the patch level, moose must choose where to forage based on the available browse species, and tree and shrub heights at different feeding stations. Younger patches can provide large quantities of high quality browse while older patches that have grown out of reach provide less browse (Schwartz 1992). Within a feed- ing station, bite size is based on the tradeoff between cropping and processing (Spalinger and Hobbs 1992). Moose need to consume about 130 g dry mass/kg body weight0.75 daily in summer and about 40 g dry mass/kg body weight0.75 daily in winter (Renecker and Hudson 1985). Using an average bite size of 1.02 g/bite (Renecker and Hudson 1986), this equates to at least 13,000 bites in summer and 4,000 bites in winter for a 454 kg (1000 lb) moose. Winter consumption may be up to 50% higher depending on browse availability and species composition (Hjeljord et al. 1994). This large demand for forage forces moose to move between patches and feeding stations in order to consume enough biomass. Browse availability and bite size have been measured by following moose or moose tracks in snow and counting the num- ber of available twigs per species, the num- ber of bites per species, and measuring diameter at point of browsing, dry mass, and twig length (Risenhoover 1987, Shipley et al. 1998). Locations of moose were found via radio telemetry (Risenhoover 1987, Hjeljord et al. 1990) or finding a track crossing a road (Shipley et al. 1998). These methods were largely opportunistic and data collection was either clumped (location 3Present Address: Hibbing Community College, 1515 25th St E, Hibbing, MN 55746 107 every hour for 2 days) or spread temporally (1–2 tracks weekly). Another typical method is to measure browse availability in plots along randomly placed straight transects instead of following moose foraging paths. This provides an esti- mate of absolute browse density in a patch, rather than an estimate of browse availability encountered by moose. We measured inten- sively used feeding patches with 3 different protocols and a randomly placed straight tran- sect in northeastern Minnesota. Our new protocol (the large feeding station method) attempted to simulate how a moose browses, which we contrasted with measurements along the foraging path and with absolute browse density. STUDY AREA This study was conducted in north- eastern Minnesota where moose were previ- ously collared for a VHF telemetry study (Fig. 1) (Lenarz et al. 2011). These forests transition between the Canadian boreal and northern hardwood forests and experience a continental climate with short warm sum- mers and severe winters (Heinselman 1996). Most was part of the Superior Nation- al Forest with the remaining either state, county, tribal, or industrial forest land Fig. 1. The study area was within the Superior National Forest in northeastern Minnesota. Each black dot represents one measured foraging path in winter and a dark gray dot represents a summer foraging path. 108 NOVEL BROWSE SURVEYS – PORTINGA AND MOEN ALCES VOL. 51, 2015 (Lenarz et al. 2010, Moen et al. 2011). More specific details are provided in the Minne- sota Moose Research and Management Plan (MNDNR 2011). METHODS Regressions and Estimating Bite Mass Summer leaves were collected between July and September 2012, and winter twigs between January and April 2012 and 2013; twigs from both years were combined in the regression analyses. We clipped (stan- dard garden clippers) browsed (∼3 cm below the browse point) and unbrowsed twigs of all browse species (Table 1). Samples were bagged and labeled with the location, date, and species. All browsed and unbrowsed twigs and leaves were stored at 2–3 °C prior to measurements. These twigs were used to develop diameter-biomass regressions for each season (Telfer 1969). In summer we collected stripped twigs of each species which we clipped directly above the first unbrowsed petiole. A winter bite was equal to the twig biomass and a summer bite the leaf biomass from one twig, both with cur- rent annual growth >5 cm. On browsed twigs we measured (nearest 0.01 mm) the diameter at point of browsing and on unbrowsed twigs the simulated diam- eter at point of browsing. In summer, the simulated point of browsing was the diam- eter underneath the last stripped petiole. The wet weight of winter twigs and stripped summer leaves was weighed to the nearest 0.01 g. After weighing the wet mass of leaves, they were placed in the same bag with the corresponding twig. All unbrowsed Table 1. The common and scientific names for each potential browse species in northeastern Minnesota and seasons in which each species is consumed. “Rare” species make up <1% of the diet at large feeding station paths. “Not Browsed” species were not consumed along the foraging paths. Common Name Scientific Name Winter Summer Balsam fir Abies balsamea Common Not Browsed Red maple Acer rubrum Common Common Mountain maple Acer spicatum Common Common Alder Alnus rugosa Rare Rare Juneberry Amelanchier spp. Common Common Paper birch Betula papyrifera Common Common Bog birch Betula pumila Not Browsed Rare Red-osier dogwood Cornus stolonifera Common Rare Hazel Corylus cornuta Common Rare Black ash Fraxinus niger Not Browsed Rare White pine Pinus strobus Rare Rare Balsam poplar Populus balsamifera Rare Rare Quaking aspen Populus tremuloides Common Common Pin cherry Prunus pennsylvanicus Common Common Choke cherry Prunus virginianus Common Common Oak Quercus spp. Not Browsed Rare Willow Salix spp. Common Common Elderberry Sambucus pubens Not Browsed Rare Mountain ash Sorbus decora Rare Common ALCES VOL. 51, 2015 PORTINGA AND MOEN – NOVEL BROWSE SURVEYS 109 twigs in both seasons were stored in la- beled bags. All unbrowsed summer and winter twigs were dried at 60 °C for 48 h in a drying oven. Dried twigs in winter and dried leaves in summer were stored at room temperature before being weighed to the nearest 0.01 g. Most winter twigs (74%) and summer leaves (90%) were measured within 5 days of re- moval from the drying oven; the remainder was measured 6–9 days later. GPS Collars We captured adult moose in February and early March 2011 by darting them from helicopters. GPS radio-collars (Sirtrack Ltd. and Lotek Wireless) fitted to each moose were programmed to transmit a location every 20 min. Animal capture and handling protocols met the guidelines recommended by the American Society of Mammalogists (Sikes et al. 2011) and were approved by University of Minnesota and National Park Service Animal Care and Use committees (#0912A75532). Measuring Browse Availability Summer browse availability was mea- sured between 25 July and 14 September 2012, and winter browse availability be- tween 3 January and 22 March 2013. Browse availability was measured at the patch scale which we identified from the GPS locations – patches had a concentrated number of moose locations. We used a handheld Gar- min GPS to reach our pre-identified patches and then searched for a feeding station to identify a foraging path. A feeding station was defined as a plant or clump of plants with browsed twigs that were accessible when the forefeet of a moose are stationary (Goddard 1968, Novellie 1978, Senft et al. 1987). A foraging path was defined as a trail of feeding stations within a patch. Summer for- aging paths were measured 1 to 15 days after the moose departed, and winter foraging paths were measured 3 to 17 days after de- parture. Patches were considered accessible if they were on public land and we could ac- cess them by walking <2 km on a trail and/or <550 m from a trail. We measured winter patches containing 29 foraging paths from 8 moose (6F, 2M), and summer patches con- taining 31 foraging paths from 7 moose (5F, 2M). We defined a large feeding station as a location that appeared to have ≥10 bites. At all sites we measured browse under 4 differ- ent protocols to produce 4 foraging path types: 1) large feeding stations along the for- aging path, 2) random plots along the for- aging path, 3) random feeding stations along the foraging path, and 4) plots along a straight transect through the area contain- ing the foraging path. Each path type con- sisted of 10 measurement plots. Large feeding station plots — The first large feeding station encountered was the first plot of the site and marked as a way- point on the handheld GPS. The plot or feed- ing station to be measured was a half circle with radius of 99.1 cm (39 in), with the cen- ter of the back side (straight line diameter) held at the approximate place where the moose stood. Tracks in winter, other sign in either season, or placement of bites relative to open space were also used to determine where the moose stood and the direction it faced. At each large feeding station we counted the unbrowsed and browsed twigs of each browse species between 0.5 and 3 m above the ground (Table 1; Shipley et al. 1998). Each cut-off twig was considered a bite. Although an occasional large feeding station had <10 bites, we included it as a large feeding station because the observer estimated it had at least 10 bites. This only occurred at 10 of 290 (3%) large feeding sta- tions in winter and 36 of 297 (12%) in summer. 110 NOVEL BROWSE SURVEYS – PORTINGA AND MOEN ALCES VOL. 51, 2015 We established the foraging path type from the first large feeding station by follow- ing tracks and browsing sign to the next large feeding station, marked it as the second waypoint on the GPS, and repeated the mea- surements (Fig. 2). Plots could not overlap and this process continued until 10 large feeding stations had been measured on the foraging path. Random plots on the foraging path — We created the random plot path type by stopping along the foraging path and repeat- ing our browse measurements in random plots. A list of random distances between 5 and 14 m was generated using Microsoft Excel, and in the field we established the random plots using these distances in the GPS “find” feature (Fig. 2). Random feeding stations — If a ran- dom plot had been browsed (evident bites), then that random plot was also defined as a random feeding station. If no browsed bites were in the random plot, we followed the foraging path to the next browsed twig (even if only one bite) and this became the location of the next random feeding station (Fig. 2), eventually creating the random feeding path type. Straight transect plots — After com- pleting the large feeding station, random plot, and random feeding station measure- ments, we established a straight line transect that returned to the first plot. Along this tran- sect we stopped at random distances between 5 and 14 m until 10 plots were measured. If we reached the first large feeding station plot before completing 10 plots, we length- ened the transect. If, however, the cover type changed past the first plot and <10 plots were measured, we established a new tran- sect in a random direction within the same cover type; 10 of 29 straight transects were angled in winter and 15 of 31 in summer. 3 1 2 8 9 54 7 6 = Large Feeding Station (≥ 10 bites) = Random Plot = Random Feeding Station (≥ 1 bite) Fig. 2. A diagram of how we measured a foraging path. Plot 1 is a large feeding station plot with ≥10 bites. Plot 2 is a random plot. Because Plot 2 did not have any bites taken we stop at the next bite which becomes Plot 3, a random feeding station plot. Plot 4 is the second large feeding station plot. Plot 5 is the second random plot with 1–9 bites, so it is also the second random feeding station plot. Plot 6 is the third large feeding station plot. Plot 7 is the third random plot that had ≥10 bites, so it is also the third random feeding station plot and the fourth large feeding station plot. Plot 8 is the fourth random plot. Plot 9 is the fourth random feeding station that had ≥10 bites, so it is also the fifth large feeding station plot. We continued until there were 10 plots of each type. ALCES VOL. 51, 2015 PORTINGA AND MOEN – NOVEL BROWSE SURVEYS 111 Some cover types had little available browse making the foraging path difficult to follow in summer when 10 of 30 foraging paths had <10 plots in all path types. If no bites were found within 20 m of the previous feeding station when moving forward, we assumed the moose stopped foraging. Effec- tively this meant that there were <10 large feeding stations, random feeding stations, and/or random plots in that foraging path. Snow tracking in winter allowed us to more easily identify the foraging path; thus, 10 plots in all path types were measured in 28 of 30 foraging paths. Canopy cover was measured 3 times with a densiometer (every 8th plot) to pro- duce an average value in each patch. Twigs collected from sites with 0–50% canopy closure were considered grown in open can- opy, and twigs from sites with 70–100% can- opy closure were considered grown in closed canopy. Twigs from sites with 51–69% can- opy cover were not used in the regressions or bite size summary statistics. Statistical Analyses Biomass-diameter at point of browsing regressions, ANOVAs on browse density, Fig. 3. The percent of random feeding stations measured in each size category (line) and the percent of bites consumed at all feeding stations of a given size category (bar) in winter and summer. The dashed line separates the small feeding stations (≤9 bites) from the large feeding stations. In winter, 57% of the random feeding stations were considered large but they accounted for 86% of the consumed bites. In summer, 49% of the random feeding stations were considered large but they accounted for 82% of the consumed bites. 112 NOVEL BROWSE SURVEYS – PORTINGA AND MOEN ALCES VOL. 51, 2015 Kruskal-Wallis comparisons of diet, Pearson χ2 Goodness of Fit tests, and Bonferroni Z-tests were all performed in Jmp 10.0. Significance level was set at P = 0.05. Regressions — Simulated diameters at point of browsing and dry masses of twigs from the unbrowsed winter twigs were log10 transformed and used to make 2 sepa- rate diameter-biomass regressions for each of the main browse species. The first regression used twigs grown in open canopy (0–50% shaded) and the second twigs from closed can- opy (70–100% shaded). Similarly, 2 summer regressions were made using leaf dry mass of each browse species. The raw data are found in Ward (2014) and only results are presented here. Statistics on bite size diameter and bite mass were calculated for each species. A t-test was used to test for statistical differ- ences between the average diameter at point of browsing in open and closed canopy in both seasons for each species. Available browse density — Browse density was estimated as twig counts and as biomass. To obtain the total number of avail- able twigs per path, we added the number of available twigs and the number of browsed bites. We estimated the total biomass origi- nally available (browsed or unbrowsed) along a foraging path by multiplying the number of twigs of a given species by the average bio- mass of one bite of that species. For foraging paths in 0–50% shade, we used the average biomass values from open canopy regres- sions. Likewise, we used the average bio- mass values from closed canopy regressions for foraging paths in 51–100% shade. Al- though the closed canopy regressions were developed with twigs grown in 70–100% shaded areas, we felt the foraging paths in 51–69% shade were better classified as closed canopy than open canopy. Balsam fir was not included in summer browse density estimates because it is not typically part of the summer diet. Available and consumed browse density along each of the 4 path types were esti- mated using twig counts and biomass in both seasons. The length of each path was calculated by measuring the length of a line passing through all of the plots of each path type. The area of the foraging path was considered twice this distance to repre- sent the ability of moose to browse either side of the foraging path. To calculate browse density we divided the twig count (available or consumed) by the area of the foraging path. These same calculations were made using biomass and twig counts. The browse density on large feeding station paths was compared with those on the ran- dom feeding, random, and straight transect paths using an ANOVA of the log trans- formed data. Diet composition — Diet composition was calculated for each moose on the 4 path types in both seasons. We made a weighted average of those diet compositions to estimate diet composition for all moose on each path type in winter and summer. Spe- cies were considered rare when they made up <1% of the average diet (Shipley et al. 1998) at large feeding station paths. The per- centage of the diet consisting of rare species is reported in the tables (but not text) to illus- trate how a few individual moose consumed many bites of rare species. Each individual diet had at least one browse species not identified on the foraging paths. Because these data were not normally distributed and no transformation could cor- rect this skewedness, we used a Kruskal- Wallis test to test for significant differences between diet composition on the 4 path types. A Kruskal-Wallis test was also used to test for differences between each individ- ual diet. Browse species selection — We also determined the selection for each browse species from a combined average of all moose and for each individual using the ALCES VOL. 51, 2015 PORTINGA AND MOEN – NOVEL BROWSE SURVEYS 113 data from large feeding station paths. A Pearson χ2 Goodness of Fit test and a Bonferroni Z-test were performed on the availability and use of all browse species for all moose combined and each individual moose (Neu et al. 1974). A species was con- sidered “positively selected”, “negatively selected”, or neither if there was a signifi- cantly larger, smaller, or equal proportion of browsed versus available twigs. RESULTS Regressions All of the twig diameter – biomass regressions had slopes significantly different from zero. The slopes ranged from 0.58– 2.80 in winter and 0.45–2.07 in summer. In winter, 75% of the regressions had an R2 >0.60, and in summer 43% had an R2 >0.60. There was no consistent pattern between the open canopy or closed canopy regression slopes being larger or smaller (Ward 2014). Bite Size Across all species in winter, the mean diameter at point of browsing was 3.0 ± 0.02 mm in open canopy (range = 0.5–9.0 mm) and 3.1 ± 0.1 mm in closed canopy (range = 0.2–8.4 mm) (Table 2). In summer, the mean across species was 2.3 ± 0.02 mm in open canopy (range = 0.02–11.1 mm) and 2.4 ± 0.04 mm in closed canopy (range = 0.2–6.1 mm) (Table 3). Using the regressions found in Ward (2014), we calculated the average biomass consumed per bite for each browse species (Tables 2 and 3). In winter, pin cherry had the largest bite size (2.3 ± 1.4 g) under closed canopy and the smallest bite size under open canopy (0.4 ± 0.1 g). Mountain maple had the smallest bite size under closed canopy (0.4 ± 0.2 g). Mountain ash had the largest (1.7 ± 1.4 g) and quaking aspen the smallest bite size (0.3 ± 0.2 g) under closed canopy in summer. Bite Density at Feeding Stations One purpose of establishing the random feeding station plots was to estimate the fre- quency of feeding stations of different sizes occurring along foraging paths. In winter 57% of random feeding station plots (n = 281) had ≥10 or more bites, and in summer 49% (n = 267). In both seasons at least 80% of twig consumption on the for- aging path was from feeding stations with ≥10 bites, although moose occasionally consumed <10 bites at a station. Browse Density Total available browse density was mea- sured at 29 patches in winter and 30 patches in summer. It was significantly different among the 4 path types in both seasons using either method (winter twigs: F3, 112= 62.7, summer twigs: F3, 118 = 32.5, winter bio- mass: F3, 112 = 84.3, summer biomass: F3, 120 = 16.8, Pall < 0.0001). Likewise, density of consumed browse was also sig- nificantly different in winter and summer among the 4 path types (winter twigs: F3, 112 = 63.4, summer twigs: F3, 120 = 31.2, winter biomass: F3, 112 = 70.9, summer biomass: F3, 119 = 5.0, Pall < 0.0025). As expected, both available and consumed browse densities were highest at large feeding station paths, followed by random feeding station, random plot, and straight transect paths (Table 4). The average available browse density estimated by biomass at large feeding sta- tions was 53% higher in summer (15.2 ± 1.7 g/m2) than winter (9.9 ± 1.0 g/m2). Conversely, density estimated by twig counts was ∼2.5x larger in winter (15.2 ± 1.6 twigs/m2) than in summer (5.9 ± 0.6 twigs/m2). Large feeding station paths had ∼60% more available twigs (727 ± 3) 114 NOVEL BROWSE SURVEYS – PORTINGA AND MOEN ALCES VOL. 51, 2015 in winter than in summer (460 ± 37), where- as the available biomass was ∼2.5x larger in summer (1166 ± 88 g) than winter (471 ± 26 g). The same seasonal differences existed for consumed twigs and biomass. The dis- tance walked in winter to complete the large feeding station paths (27.6 ± 2.0 m, n = 29) was about half that in summer (50.5 ± 4.9 m, n = 31). The available and consumed browse density for each browse species was largest at large feeding station paths followed by random feeding station, random plot, and straight transect paths. The one exception (based on twig count) was that the highest browse density of hazel was found on the straight transect path in summer (when hazel is rarely consumed). Table 2. Summary statistics on browsed twigs in winter for all browse species. Open canopy indicates twigs grown in locations shaded 0–50% and closed canopy indicates twigs grown in locations shaded 70–100%. P-values indicate t-test results between the diameter at point of browsing (DPB) of each species in open and closed canopy. We did not find enough individual twigs of juneberry, paper birch, pin cherry, or willow in closed canopy to calculate reliable averages for those categories. Diameter at Point of Browsing (mm) Species Canopy Average ± SE Minimum Maximum Average Bite ± SE (g) n P Balsam fir** Open 2.7 ± 0.1 0.9 6.5 1.6 ± 0.3 82 0.002 Closed 2.2 ± 0.1 1.0 4.0 1.2 ± 0.2 50 Red maple** Open 3.5 ± 0.1 1.3 7.4 0.7 ± 0.3 125 0.009 Closed 4.1 ± 0.1 2.7 6.9 1.4 ± 0.5 27 Mountain maple* Open 2.8 ± 0.3 1.5 4.6 0.6 ± 0.3 47 0.019 Closed 2.4 ± 0.3 0.4 4.9 0.4 ± 0.2 56 Juneberry Open 2.4 ± 0.1 0.9 4.5 0.5 ± 0.1 161 0.583 Closed NA NA NA NA 8 Paper birch Open 2.7 ± 0.1 0.6 4.8 0.8 ± 0.1 188 NA Closed NA NA NA NA 7 Hazel Open 2.7 ± 0.1 1.1 5.3 0.6 ± 0.1 301 0.104 Closed 2.8 ± 0.1 1.1 4.5 0.6 ± 0.1 132 Red-osier dogwood*** Open 3.5 ± 0.1 1.5 6.1 1.1 ± 0.1 332 <0.0001 Closed 4.3 ± 0.2 2.0 6.6 1.4 ± 0.4 40 Quaking aspen Open 3.5 ± 0.1 0.9 6.8 0.9 ± 0.1 209 0.155 Closed 3.2 ± 0.1 1.0 5.7 0.7 ± 0.4 32 Pin cherry Open 2.4 ± 0.1 0.6 4.9 0.4 ± 0.1 216 NA Closed NA NA NA NA 6 Choke cherry Open 3.0 ± 0.3 1.5 4.8 0.7 ± 0.1 53 0.120 Closed 2.6 ± 0.4 0.2 4.1 0.4 ± 0.1 20 Willow Open 3.1 ± 0.1 0.5 6.4 0.9 ± 0.1 501 NA Closed1 NA NA NA NA 0 Mountain ash* Open 4.3 ± 0.1 1.6 6.8 1.3 ± 0.3 43 0.045 Closed 3.7 ± 0.1 1.2 8.4 0.7 ± 0.5 53 Combined Open 3.0 ± 0.02 0.5 9.0 NA 2388 Closed 3.1 ± 0.1 0.2 8.4 NA 454 ALCES VOL. 51, 2015 PORTINGA AND MOEN – NOVEL BROWSE SURVEYS 115 Consumption Rate The pattern of consumption rate was similar to that of consumed browse density. The proportion of consumed twigs was high- est on the large feeding station paths and declined progressively to the random feeding station, random plot, and straight transect paths. Consumption was 45% in summer and 35% in winter on the large feeding sta- tion paths. Overall, it was 23–45% on all paths except the straight transects where rates were 13% in winter and 9% in summer. Diet Composition Season — At least 70% of all bites (all moose) consumed in winter along the 4 path types consisted of hazel, paper birch, willow, and quaking aspen. The remaining 30% con- sisted of balsam fir, juneberry, mountain maple, red maple, red-osier dogwood, pin Table 3. Summary statistics on browsed twigs of all species in summer. Open canopy indicates twigs grown in locations shaded 0–50% and closed canopy indicates twigs grown in locations shaded 70–100%. P- values indicate t-test results between the diameter at point of browsing (DPB) of each species in open and closed canopy. We did not find enough individual twigs of red maple in open canopy or pin cherry, willow, or mountain ash in closed canopy to calculate reliable averages for those categories. Diameter at Point of Browsing (mm) Species Canopy Mean ± SE Minimum Maximum Mean Bite ± SE (g) n P Red maple Open NA NA NA NA 14 0.349 Closed 2.8 ± 0.2 1.3 6.0 1.4 ± 0.3 27 Mountain maple*** Open 2.3 ± 0.03 0.5 4.7 0.7 ± 0.1 675 <0.0001 Closed 3.0 ± 0.1 0.5 4.9 1.0 ± 0.1 264 Juneberry Open 1.6 ± 0.04 0.1 3.2 0.5 ± 0.04 149 0.145 Closed 2.1 ± 0.3 0.2 4.2 1.0 ± 0.4 20 Paper birch** Open 2.3 ± 0.1 0.02 5.1 0.8 ± 0.1 316 0.003 Closed 2.0 ± 0.1 0.6 3.8 0.5 ± 0.1 84 Hazel Open 1.6 ± 0.1 0.5 3.5 0.7 ± 0.04 105 0.739 Closed 1.6 ± 0.1 0.6 2.5 0.6 ± 0.1 48 Red-osier dogwood*** Open 2.9 ± 0.1 1.5 5.7 1.3 ± 0.1 41 0.001 Closed 2.1 ± 0.2 0.5 4.4 0.7 ± 0.1 26 Quaking aspen*** Open 3.1 ± 0.2 0.5 11.1 1.4 ± 0.2 169 <0.0001 Closed 1.6 ± 0.1 0.3 4.3 0.3 ± 0.2 53 Pin cherry Open 2.2 ± 0.1 0.6 4.2 0.8 ± 0.1 53 NA Closed NA NA NA NA 0 Choke cherry Open 2.2 ± 0.1 1.0 4.1 0.8 ± 0.1 44 0.085 Closed 2.0 ± 0.1 0.8 3.9 0.8 ± 0.1 80 Willow*** Open 2.3 ± 0.1 0.5 5.5 0.9 ± 0.1 242 <0.0001 Closed NA NA NA NA 14 Mountain ash Open 4.0 ± 0.1 2.0 7.0 1.1 ± 0.1 72 0.802 Closed NA NA NA NA 7 All Species Open 2.3 ± 0.02 0.02 11.1 NA 2071 NA Closed 2.4 ± 0.04 0.2 6.1 NA 627 116 NOVEL BROWSE SURVEYS – PORTINGA AND MOEN ALCES VOL. 51, 2015 cherry, and choke cherry. Rare species were alder, mountain ash, balsam poplar, and white pine (Table 5). In summer 70% of bites consisted of mountain maple, willow, and paper birch on large feeding station, random feeding sta- tion, and random plot paths. The remaining 30% was juneberry, red maple, pin cherry, choke cherry, quaking aspen, and mountain ash. Rare species were hazel, balsam poplar, red-osier dogwood, balsam fir, alder, bog birch, black ash, oak, elderberry, and white pine. On straight transects at least 70% of consumed twigs were mountain maple, wil- low, quaking aspen, and species considered rare (Table 5). Path type — Despite the general simi- larities in diet diversity, all browse species comprised different portions of the winter diet on the 4 path types (Kruskal-Wallis, H3 > 12.3, P < 0.007) except paper birch and hazel (Kruskal-Wallis, H3 < 1.2, P > 0.60; Table 6). In summer Juneberry, quaking aspen, and mountain ash comprised different portions of the diet on all 4 path types in summer (Kruskal-Wallis, H3 > 8.1, P < 0.045; Table 5). No difference existed among the 4 path types for red maple, mountain maple, paper birch, cherry, and willow (Kruskal-Wallis, H3 < 5.7, P > 0.13). Individuals — Diets based on twigs consumed on large feeding station paths varied individually and from the pooled average (Tables 6 and 7). One winter ex- ample of this individual difference was fe- male moose 31180 that consumed 26% red maple and 50% hazel (4 paths combined) compared to the group average of 5% red maple and 26% hazel (Table 6); red maple was more available in her foraging patches. An example in summer was male moose 31190 that consumed 10% mountain maple and 61% willow (4 paths combined) com- pared to the group average of 41% mountain maple and 21% willow (Table 7). Browse Species Selection The average diet in winter (all moose combined) was different from that avail- able (v29 ¼ 3122, P < 0.0001). A Bonferroni Z-test on the combined data indicated that juneberry, red maple, mountain maple, paper birch, red-osier dogwood, and quaking aspen were eaten more than available in summer. Hazel was eaten less than available, and cherry and willow were used in proportion to availability (Table 8). Individual diets were also different from browse availability on their respective foraging paths (all moose: v2�9 � 74:6, P < 0.0001 for all moose). The average summer diet (all moose combined) was also different from available (v28 ¼ 840, P < 0.0001), as were individual diets (all moose: v2�8 � 43:9, P < 0.0001). Table 4. Available browse density and consumed browse density along four path types in summer and winter measured by twigs/m2 ± SE and biomass (g)/m2 ± SE. W = winter, S = summer. Method Season Large Feeding Station Random Feeding Station Random Plot Straight Transect Available # Twigs W 15.4 ± 1.6 2.3 ± 0.2 2.0 ± 0.2 1.4 ± 0.2 S 5.9 ± 0.6 2.0 ± 0.2 1.8 ± 0.3 1.1 ± 0.1 Biomass W 9.9 ± 1.0 1.7 ± 0.1 1.5 ± 0.1 1.0 ± 0.1 S 15.2 ± 1.7 6.8 ± 1.9 4.5 ± 0.8 2.9 ± 0.4 Consumed # Twigs W 5.3 ± 0.6 2.1 ± 0.1 0.5 ± 0.1 0.2 ± 0.03 S 2.7 ± 0.3 1.0 ± 0.3 0.4 ± 0.1 0.1 ± 0.03 Biomass W 4.0 ± 0.4 0.5 ± 0.04 0.4 ± 0.04 0.2 ± 0.02 S 6.7 ± 0.7 2.4 ± 0.4 1.0 ± 0.2 0.3 ± 0.04 ALCES VOL. 51, 2015 PORTINGA AND MOEN – NOVEL BROWSE SURVEYS 117 A Bonferroni Z-test on the combined sum- mer data indicated that red maple, mountain maple, cherry, and mountain ash were eaten more than available in summer, willow less than available, and juneberry, paper birch, and quaking aspen proportional to availability (Table 8). DISCUSSION We initially chose to measure large feeding stations (≥10 bites) because field observations indicated that these sites were common and theory (Senft et al. 1987) sup- ports the strategy of such foraging behavior. By contrasting browse density along a foraging path at large feeding stations with alternate routes, we demonstrated how moose increased effective browse density by selecting a specific foraging path. For ex- ample, moose took at least 80% of their bites at large feeding stations with ≥10 bites. The identification of large feeding stations pro- vided a fast and efficient manner to measure browse availability and consumption along presumed foraging paths, and this method can also be used to evaluate the relative qual- ity of browsed and unbrowsed patches (Ward 2014, Ward and Moen 2014) This method avoids 2 potential compli- cations associated with the straight transect Table 5. Diet composition (average percent of diet ± SE) measured on four path types. Averages and SE were weighted by moose. Rare includes species that made up <1% of the diet at large feeding station paths. 29 foraging paths were measured in winter 2013 and 31 were measured in summer 2012. Winter Species Large Feeding Station Random Feeding Station Random Plot Straight Transect Hazel 27 ± 7 26 ± 8 27 ± 9 28 ± 8 Paper birch 26 ± 7 26 ± 6 25 ± 6 18 ± 6 Willow 11 ± 5 14 ± 6 13 ± 6 11 ± 5 Quaking aspen 7 ± 3 8 ± 4 10 ± 5 13 ± 6 Juneberry 6 ± 2 5 ± 2 4 ± 1 4 ± 2 Red maple 5 ± 3 4 ± 2 5 ± 3 4 ± 4 Red-osier dogwood 5 ± 4 3 ± 3 3 ± 3 10 ± 11 Balsam fir 4 ± 2 6 ± 2 6 ± 3 2 ± 2 Mountain maple 4 ± 3 3 ± 1 2 ± 1 2 ± 1 Cherry 3 ± 1 2 ± 1 2 ± 1 2 ± 1 Rare 2 ± 2 2 ± 1 2 ± 1 5 ± 6 Summer Mountain maple 42 ± 11 45 ± 10 43 ± 11 25 ± 11 Willow 21 ± 8 21 ± 9 28 ± 11 23 ± 11 Paper birch 11 ± 3 9 ± 4 6 ± 4 6 ± 5 Cherry 9 ± 4 7 ± 4 6 ± 4 3 ± 5 Quaking aspen 8 ± 4 10 ± 3 8 ± 3 14 ± 7 Mountain ash 4 ± 2 3 ± 2 4 ± 4 0 Juneberry 2 ± 1 3 ± 2 2 ± 1 8 ± 5 Red maple 1 ± 1 0 0 7 ± 4 Rare 1 ± 0.3 1 ± 0.4 0.2 ± 0.1 10 ± 7 118 NOVEL BROWSE SURVEYS – PORTINGA AND MOEN ALCES VOL. 51, 2015 method: 1) measuring random locations, and 2) empty plots. The foraging path approach eliminates these concerns by ensuring plenti- ful data at actual foraging locations. Argu- ably, it also reflects the browse availability a moose would actually perceive. Randomly placed plots in straight transects are often empty, which would mean that many more plots would be required to accurately esti- mate the availability of patchy browse. Our method avoids empty plots, incorporates dis- tance moved between feeding stations, and Table 6. Diet composition of individual moose in winter 2013 measured by twigs consumed at large feeding station paths. There are diets for eight collared moose. 31189 and 31190 are male, the rest are females. N is the number of foraging paths measured. Rare species made up <1% of the combined moose diet at large feeding stations. Moose Number Species All Moose 31166 31174 31175 31178 31180 31182 31189 31190 Hazel 27 21 38 29 13 50 33 9 68 Paper birch 26 14 41 15 57 9 3 56 3 Willow 11 5 9 6 5 3 Quaking aspen 7 28 16 12 <1 8 2 Juneberry 6 18 1 8 9 1 Red maple 5 26 9 Red-osier dogwood 5 15 <1 38 4 Balsam fir 4 2 4 15 1 14 Mountain maple 4 16 1 1 25 Cherry 3 11 1 5 6 <1 3 6 Rare 1 2 5 1 N 29 2 2 3 3 4 2 5 3 Table 7. Diet composition of individual moose in summer 2012 measured by twigs consumed at large feeding stations only. There are diets for seven collared moose. 31189 and 31190 are male, the rest are females. N is the number of sites measured. Rare species made up <1% of the combined moose diet at large feeding stations. Moose Number Species All Moose 31166 31168 31175 31178 31180 31189 31190 Mountain maple 41 3 57 84 90 36 57 10 Willow 21 53 3 9 17 61 Paper birch 11 12 1 17 13 7 Cherry 9 8 5 3 2 24 1 3 Quaking aspen 8 4 36 1 23 2 Mountain ash 4 17 5 5 3 Juneberry 2 1 12 Red maple 1 5 Rare 1 1 1 3 N 31 3 2 3 3 3 6 4 ALCES VOL. 51, 2015 PORTINGA AND MOEN – NOVEL BROWSE SURVEYS 119 provides an estimate of effective browse density. A challenge to simulating foraging decision rules when following a foraging path is that humans find large feeding sta- tions by sight, but moose likely use other senses as well. Diet composition was statistically differ- ent among seasons and path types. The aver- age combined diet in both winter and summer was best categorized as generalist because one genus did not account for >60% of the diet (Shipley 2010). The two primary browsed species were hazel and paper birch in winter and mountain maple and willow in summer, hence, moose may forage in different areas in winter and sum- mer. For example, available browse density estimated by twig counts was higher in win- ter than in summer, with hazel consumed commonly in winter but rarely in summer. Use of GPS locations may help distinguish seasonal differences in foraging locations and browse species availability. The diet composition was similar to that measured >3 decades previously in the re- gion (Peek et al. 1976). The top 5 summer species (percent of diet) were the same in both studies: mountain maple, willow, paper birch, cherry, and quaking aspen. Mountain maple was ranked first in our study and fifth by Peek et al. (1976), and quaking aspen had the opposite rankings. Hazel, willow, and quaking aspen were 3 of the top 5 winter species in both studies. One difference was that paper birch and juneberry were included in our top 5, whereas Peek et al. (1976) had balsam fir and red-osier dogwood. During both seasons the primary species consumed Table 8. Browse species selection in both seasons when data from all moose was combined. If the moose were simply browsing at random, we would expect the 95% confidence interval of the percent browsed to contain the percent available at large feeding stations. Season Species Percent Available at Large Feeding Stations 95% Confidence Interval of Percent Browsed at Large Feeding Stations Selection Winter Juneberry 4.7 5.1 ≤ – ≥ 6.8 + Red maple 3.3 3.8 ≤ – ≥ 5.3 + Mountain maple 2.7 4.0 ≤ – ≥ 5.5 + Paper birch 19.3 24.7 ≤ – ≥ 27.9 + Red-osier dogwood 2.1 3.3 ≤ – ≥ 4.8 + Quaking aspen 5.6 5.8 ≤ – ≥ 7.6 + Cherry 3.0 2.7 ≤ – ≥ 4.0 0 Willow 11.9 11.2 ≤ – ≥ 13.5 0 Balsam fir 9.0 2.8 ≤ – ≥ 4.1 − Hazel 36.8 26.3 ≤ – ≥ 29.5 − Summer Red maple 0.5 0.6 ≤ – ≥ 1.3 + Mountain maple 27.6 34.6 ≤ – ≥ 38.2 + Cherry 7.2 8.3 ≤ – ≥ 10.5 + Mountain ash 4.2 8.6 ≤ – ≥ 10.8 + Juneberry 3.3 2.2 ≤ – ≥ 3.4 0 Paper birch 10.4 9.8 ≤ – ≥ 12.1 0 Quaking aspen 8.1 6.1 ≤ – ≥ 8.1 0 Willow 28.6 18.9 ≤ – ≥ 21.9 − 120 NOVEL BROWSE SURVEYS – PORTINGA AND MOEN ALCES VOL. 51, 2015 were consistent regardless of path type. Be- cause more twigs were counted on the large feeding station paths, they probably pro- vided the better estimate of diet and species consumption rates. This study was unique because we col- lected data from individual free-ranging moose by using their GPS locations to iden- tify their foraging paths shortly after use. Presumably each moose selected browse based on availability within the patch they occupied. Individual consumption differ- ences occurred in both winter and summer, and though previous studies have not pro- vided for analysis and comparison of indi- vidual diet selection, individual differences in habitat selection by moose were documen- ted in British Columbia (Gillingham and Parker 2010). Pooling the data from many foraging paths identified the generalized sea- sonal diets and the most important browse species in this region, and concurred with previous research. It also identified individ- ual diet variation which suggests that moose adapt their diet based on the local compo- sition and availability of browse species. We were able to simulate how a moose browsed in a patch using the large feeding station method. There was some subjectivity in choosing which large feeding station was closest (consecutive) when establishing the foraging path; however, a moose would face the same choice. Contrasting browse measurements between simulated and actual foraging paths in the same patch would pro- vide a good evaluation of our approach and potential differences. We offer that incorpor- ating large feeding stations and the distance between adjacent large feeding stations is an efficient method to estimate browse avail- ability at the patch level. ACKNOWLEDGEMENTS We would like to thank the EPA Great Lakes Restoration Initiative and the Minne- sota Environment and Natural Resources Trust Fund (ENRTF) for funding and N. Bogyo of the 1854 Treaty Authority for winter field work assistance. REFERENCES GILLINGHAM, M. P., and K. L. PARKER. 2010. Differential habitat selection by moose and elk in the Besa-Prophet area of north- ern British Columbia. Alces 44: 41–63. GODDARD, J. 1968. Food preferences of two black rhinoceros populations. African Journal of Ecology 6: 1–18. HEINSELMAN, M. 1996. The Boundary Waters Wilderness Ecosystem. University of Minnesota Press, Minneapolis, Minne- sota, USA. HJELJORD, Ø., N. HOVIK, and H. B. PEDERSEN. 1990. Choice of feeding sites by moose during summer, the influence of forest structure and plant phenology. Holarctic Ecology 13: 281–292. ———, B. E. SAETHER, and R. ANDERSEN. 1994. Estimating energy intake of free- ranging moose cows and calves through collection of feces. Canadian Journal of Zoology 72: 1409–1415. LENARZ, M. S. 2011. 2011 Aerial Moose Sur- vey. Minnesota Department of Natural Resources, St. Paul, Minnesota, USA. LENARZ, M. S., J. FIEBERG, M. W. SCHRAGE, and A. J. EDWARDS. 2010. Living on the edge: viability of moose in northeastern Minnesota. Journal of Wildlife Manage- ment 74: 1013–1023. (MNDNR) Minnesota Department of Natural Resources. 2011. Minnesota Moose Research and Management Plan. Minne- sota Department of Natural Resources, St. Paul, Minnesota, USA. MOEN, R. A., R. PETERSON, S. WINDELS, L. FRELICH, D. BECKER, and M. JOHNSON. 2011. Minnesota Moose Status: Progress on Moose Advisory Committee Recom- mendations. NRRI Technical Report No. NRRI/TR-2011/41. Natural Resources Research Institute, Duluth, Minnesota, USA. ALCES VOL. 51, 2015 PORTINGA AND MOEN – NOVEL BROWSE SURVEYS 121 NOVELLIE, P. A. 1978. Comparison of the for- aging strategies of blesbok and springbok on the Transvaal Highveld. South African Journal of Wildlife Research 8: 137–144. NEU, C. W., C. R. BYERS, and J. M. PEEK. 1974. A technique for analysis of utiliza- tion-availability data. Journal of Wildlife Management 38: 541–545. PEEK, J. M., D. L. URICH, and R. J. MACKIE. 1976. Moose habitat selection and rela- tionships to forest management in north- eastern Minnesota. Wildlife Monographs 48: 3–65. RENECKER, L. A., and R. J. HUDSON. 1985. Estimation of dry matter intake of free- ranging moose. Journal of Wildlife Management 49: 785–792. ———, and ———. 1986. Seasonal for- aging rates of free-ranging moose. Jour- nal of Wildlife Management 50: 143–147. RISENHOOVER, K. L. 1987. Wintering foraging strategies of moose in subarctic and bor- eal forest habitats. Ph. D. Thesis, Mich- igan Technical University, Houghton, Michigan, USA. SCHWARTZ, C. C. 1992. Physiological and nu- tritional adaptations of moose to northern environments. Alces Supplement 1: 139–155. SENFT, R. L., M. B. COUGHENOUR, D. W. BAILEY, L. R. RITTENHOUSE, O. E. SALA, and D. M. SWIFT. 1987. Large herbivore foraging and ecological hierarchies. BioScience 37: 789–799. SHIPLEY, L. A. 2010. Fifty years of food and foraging in moose: lessons in ecology from a model herbivore. Alces 46: 1–13. ———, S. BLOMQUIST, and K. DANELL. 1998. Diet choices made by free-ranging moose in northern Sweden in relation to plant distribution, chemistry, and morphology. Canadian Journal of Zoology 76: 1722–1733. SIKES, R. S., and W. L. GANNON. 2011. The Animal Care and Use Committee of the American Society of Mammalogists. (2011). Journal of Mammalogy 92: 235–253. SPALINGER, D. E., and N. T. HOBBS. 1992. Mechanisms of foraging in mammalian herbivores: new models of functional re- sponse. The American Naturalist 140: 325–348. TELFER, E. S. 1969. Twig weight-diameter relationships for browse species. Journal of Wildlife Management 33: 917–921. WARD, R. L. 2014. Browse availability, bite size, and effects of stand age on species composition and browse density for moose in northeastern Minnesota. M.S. Thesis, University of Minnesota, Minne- apolis, Minnesota, USA. ———, and R. A. Moen. 2014. Effects of stand age on species composition and browse density in northeastern Minne- sota. NRRI Technical Report No. 2014– 36. Natural Resources Research Institute, Duluth, Minnesota, USA. 122 NOVEL BROWSE SURVEYS – PORTINGA AND MOEN ALCES VOL. 51, 2015 A NOVEL METHOD OF PERFORMING MOOSE BROWSE SURVEYS STUDY AREA METHODS Regressions and Estimating Bite Mass GPS Collars Measuring Browse Availability Statistical Analyses RESULTS Regressions Bite Size Bite Density at Feeding Stations Browse Density Consumption Rate Diet Composition Browse Species Selection DISCUSSION ACKNOWLEDGEMENTS REFERENCES