1 MOOSE USE OF THE MOUNT MCALLISTER BURN IN NORTH-CENTRAL BRITISH COLUMBIA: INFLUENCE OF BURN SEVERITY AND SOIL MOISTURE Kayla G. McNay1, R. Scott McNay2, Krista Sittler2, and Roy V. Rea1 1Natural Resources and Environmental Studies Institute, University of Northern British Columbia, 3333 University Way, Prince George, British Columbia V2N 4Z9, Canada; 2Wildlife Infometrics Inc., #3-220 Mackenzie Blvd., Mackenzie, British Columbia V2O 2P0, Canada ABSTRACT: The influence of recent wildfires in British Columbia (BC) on moose habitat and its use by moose are understudied, as are prescribed burning strategies that can be used to enhance moose habitat. Our objective was to investigate how 3 classes of fire severity (high, medium, low) interact with 3 soil moisture regimes (hydric, mesic, xeric) in determining how moose use post-fire habitat. In north-central BC, we studied moose use at 2 different spatial levels in the 5-year-old, 26,500 ha Mt. McAllister burn. At the site level, we estimated the density of fecal pellet groups and the percent of plants browsed by moose within plots of varying burn severity and soil moisture. At the landscape level, we investigated use from GPS locations of 7 radio-collared female moose at 3 orders of selec- tion: we compared: 1) randomly distributed locations within the home range to randomly distributed locations throughout the entire burn (2nd order of selection); 2) use locations to randomly distributed potential locations within the home range (3rd order of selection); and daily use locations with poten- tial movement locations (4th order of selection). At the site level, moose used areas of low/medium fire severity and hydric soil moisture. At the landscape level, moose preferred areas of medium fire sever- ity at the daily order, and low/medium fire severity at both the home range and burn orders of selec- tion. Our findings highlight that moose use of post-fire habitat varied by spatial scale and by order of selection and that researchers assessing use of burns by moose should consider multiple levels of investigation. Prescribed burning to enhance moose habitat should focus on low/medium fire severity at sites with mesic soil moisture. ALCES VOL. 57:1–22 (2021) Key words: Alces alces, burn, burn severity, forest fire, habitat, habitat management, moose, selection, soil moisture Moose (Alces alces) are an important component of the predator-prey dynamic in northern British Columbia (BC) as they con- stitute a large portion of prey biomass, mak- ing them an important keystone species (Gillingham and Parker 2008). Moose are also a valued resource for First Nations, for sustenance and ceremonial purposes and an important part of the big-game harvest for resident hunters and guide outfitters (Gorley 2016). Recent surveys performed by the Ministry of Forests, Lands, Natural Resource, Operations and Rural Development (MFLNRORD) estimate the current (2018) moose population in BC at 120,000–205,000, a decline of ~25,000–35,000 since 2014 (MFLNRORD 2018). This decline is con- cerning to the public, land and resource users, and natural resource managers; as a result, population monitoring is part of the strategic approach in management and deci- sion-making (Kuzyk et al. 2018a, Sittler and McNay 2018). Reasons for the decline are complex, vary regionally, and often FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 2 interrelated and include altered habitat due to logging and mining, predation, and climate change (Gorley 2016, Kuzyk et al. 2018b, Sittler 2020). Wildfire is a natural disturbance that alters moose habitat in the short and long- term by resetting vegetative communities to an early successional stage (Johnson and Hale 2002). Wildfires in BC are becoming more frequent, larger, and often more severe, pre- sumably due to increased temperatures, drier conditions, and a record high Build Up Index or the total amount of fuel available for com- bustion (BCWS 2018). With the exception of moose densities remaining unchanged prior to and after the Plateau Fire in the Cariboo Region (MFLNRORD 2018), there is mini- mal information about the influence of larger and more severe wildfires on population demographics and habitat use by moose in the Province. Prescribed fire was used frequently by First Nations as a management tool to clear land for agriculture and manipulate move- ment of wildlife into desirable hunting loca- tions (Johnson and Hale 2002); more recently, fire is used to improve forage con- dition for free-ranging cattle (Johnson and Hale 2002, Dixon et al. 2018). The same principle is now applied to enhance avail- ability and abundance of moose forage (Goddard 2011). Many factors influence post-fire revegetation including pre-fire spe- cies composition, timing of the fire (i.e., time-since fire and season), fire intensity and frequency, and size of the burn (MacCracken and Viereck 1990). There has been little consideration of how best to conduct prescribed burns to achieve efficiently the most desirable effects for moose. For example, the site series char- acteristics of soil moisture and soil nutrient regime influence plant species composition and abundance within a site (Meidinger and Pojar 1991). Pre-fire vegetative composition, site series, and season are interrelated and must be considered together to optimize man- agement objectives. Fire severity is more influenced directly by fire intensity than indi- rectly by soil moisture (Keeley 2009), although post-fire vegetative regrowth will be influenced by both. We examined how fire severity interacts with soil moisture to influence habitat use by moose within the Mount McAllister burn in northeastern BC. Using moose fecal pellet and browse data at the site level and GPS locations of collared female moose at the landscape level, we tested a null hypothesis that moose equally use sites of varying burn severity, soil moisture, and browse composi- tion. Our objectives were to determine habi- tat use response to the interaction between burn severity and soil moisture, and provide related management information about the value of post-fire habitats for moose and the value of prescribed burning. STUDY AREA We studied the Mt. McAllister burn (hence forth “the burn”) because of its vary- ing classes of burn severity and the avail- ability of location data from GPS-collared moose using the burn. Located 56 km west of Chetwynd, BC, the fire started on the southeast-facing slope of Mt. McAllister and was ~26,280.8 ha in size (Fig. 1). According to the British Columbia Wildfire Service, the fire was detected 13 July 2014 by tanker groups actioning nearby fires and was caused by lightning. We assumed that natural vege- tative succession was sufficient to produce an abundance of moose forage, and therefore provide useful information about prescribed burn management for moose. The major drainage is the Peace River, located 30 km north of the burn, with multiple creeks (Dowling Creek, Johnson Creek, McAllister Creek, and Gething Creek) sur- rounding and running through the burn and ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 3 draining into the Peace River; elevation ranges from 502 to 2,309 m. Biogeoclimatic zones within the burn are Engelmann Spruce- Subalpine Fir (ESSFmv2) at higher, Sub- boreal Spruce (SBSwk2) at middle, and Boreal White and Black Spruce (BWBSwk2) at lower elevation (Meidinger and Pojar 1991). Prior to the fire, forests were composed of a mixed overstory canopy of white spruce (Picea glauca) and subalpine fir (Abies lasiocarpa), and an understory of willow (Salix spp.), paper birch (Betula papyrifera), and trembling aspen (Populus tremuloides). Throughout the study area there were a series of clearcuts of varying age planted with lodgepole pine (Pinus con- torta subsp. latifolia) and white spruce. Fig. 1. Mt. McAllister burn study area in north-central British Columbia with 1 km2 sample blocks (grey grid). Sampled blocks are outlined in bold and corresponding sample plots within those blocks are shown as dots. FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 4 The nearby moose population density was estimated as 0.44 moose/km2 in 2018 (Sittler and McNay 2018), a small decline from 0.50 moose/km2 in 2013 (Lirette 2013). Other ungulates that occupy the area in and around the burn include caribou (Rangifer tarandus), elk (Cervus canaden- sis), white-tailed deer (Odocoileus virgin- ianus), and mule deer (O. hemionous) (Shackleton 2013). Potential predators of these ungulates are grey wolf (Canis lupus), black bear (Ursus americanus), grizzly bear (U. arctos), cougar (Puma concolor), and wolverine (Gulo gulo) (Hatler et al. 2018). METHODS Site Level We chose to use random plot-based sam- pling over investigating individual locations of radio-collared moose (see below) to gain more geographic precision than remote telemetry. To examine the use of burned sites by moose, we generated a layer of potential sample blocks from a 1km2 grid in ArcGIS (Fig. 1). Within that layer we assessed: 1) variation in burn severity using a post-burn image captured by Landsat 8 Operational Land Imager (OLI), and 2) frequency of use by moose using imported locations from GPS-collared cow moose (see below). We identified blocks with a range of burn sever- ity (low, medium, high) and high use by moose for potential sampling; no unburned areas were surveyed or used as controls. We then selected 5 sample blocks that were road and helicopter accessible and that fit within budgetary and logistical constraints of the project. A power-of-test conducted after the first year led us to increase the number of blocks in year 2 to achieve a more robust sample (n = 8, Fig. 1). Within those blocks, we identified 50 sample plots (radius = 3.99 m; Saether and Andersen 1990) distributed randomly and spaced ≥ 20 m apart using the random points in polygons tool in ArcGIS (Fig. 2). The random sample plots were the basis for addressing our objectives at the site level by assessing the response variable (use by moose) and the independent predictor variables (burn severity and soil moisture) within each plot. We visually classified burn severity (low, medium, high) of each plot using a modified composite Burn Index (Key and Benson 2003) by estimating the presence or absence of leaves, needles, and branches to assess the extent of burning in the upper and lower canopy trees. We also estimated the amount of organic material remaining below the newly distributed leaf litter. To estimate relative soil moisture, we dug a 30 cm radius x 60 cm deep soil pit at plot center to assess soil moisture (hydric, mesic, xeric), texture, and nutrient regime (Weil and Brady 2017). To estimate use by moose at the site level, we measured the density of fecal pellet groups in each plot and the proportion of plants browsed by moose; both are reliable indicators of habitat use (Mansson et al. 2007, 2011). We searched for pellet groups (≥ 5 individual pellets) in the main 3.99 m plot and within 4 subplots of equal size established in each cardinal direction imme- diately adjacent to the main plot boundary. Pellet density (groups/m2) equaled the num- ber of pellet groups divided by subplot area. To correct for visibility bias associated with obstructions (e.g., downed trees), we cor- rected proportionally for the estimated amount of obstructed area. We measured 4 primary (preferred) for- age species in each plot: willow, trembling aspen, subalpine fir, and paper birch (Goddard 2003). We did not measure the amount of shoots browsed on individual plants; rather, plants were classified as browsed or not. To estimate percentage browsed, we visually assessed the propor- tion of plants browsed and multiplied it by ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 5 the proportion available for each preferred forage species: Plants browsed (%) PB PC | 100%i ii 1 4∑ )(= × × = (Eq. 1) where i was each of 4 preferred browse spe- cies, PBi was a visual estimate of the propor- tion of total amount of species i that showed signs of browsing, and PCi was a visual estimate of the proportion of the three- dimensional plot space occupied by species i. For example, if there were 10 shrubs in a plot and 5 were browsed, the percentage browsed was 50%. Collared Moose We assessed location data (collected once daily at 0900 hr, February 2018 – January 2019) from 42 GPS-collared female moose that were part of a larger project examining limiting factors in the area (Sittler and McNay 2017). We identified animals using the burn by estimating the combined seasonal and annual home ranges of individual moose with a 95% kernel den- sity estimator (Seaman et al. 1999, Healy et al. 2018) in ArcMap (ESRI 2015). Although ranges of 10 moose overlapped the burn, 3 animals with minimal use (< 20% loca- tions) were discarded; sample size was 7 animals with 874 locations (mean = 125/ animal; SD = 75; Table 1). Five of 7 used the burn in all seasons and were considered resident; 2 were migratory using the burn only during winter. Landscape Level We considered all area within the fire perimeter as having been burned. We used Fig. 2. McAllister burn perimeter with 6 moose home ranges overlapping the burn. The image on the left depicts the delta normalized burn ratio used to determine the level of fire severity, and the image on the right depicts the SAGA wetness index used to determine level of soil moisture. FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 6 the GPS location data to determine habitat use by moose within the burn at three stan- dard orders of selection (sensu Johnson 1980); hence, referring to this as the land- scape-level focus. First, we used second- order selection to compare home ranges within the burn to the entire burn using points within the home range and random points placed every 2 ha; hereafter, burn order. Next, we used third-order selection to compare use locations to available locations within the home range randomly placed every 2 ha; hereafter, home range order. Finally, we assessed habitat use with daily movements; hereafter, fourth-order selec- tion or daily order. Each animal’s potential daily movement was estimated following methods of Sittler (2020) by placing a cir- cular buffer around each daily use location in a GIS. A season-specific buffer radius was determined using the 95th-percentile of distances traveled by individuals between consecutive daily locations within each sea- son. We then selected 5 random locations within each buffered use location (4370 points total). Seasons were defined in Sittler (2020) with consideration of normal behav- ioral changes (Gillingham and Parker 2008, Heard et al. 2013, Scheideman 2019): calv- ing (16 Apr–15 Jun), summer (16 Jun–15 Aug), fall (16 Aug–31 Oct), winter (1 Nov– 31 Jan), and late winter (1 Feb–15 Apr). We characterized all use and random locations for each analysis with estimates of burn severity and soil moisture. To estimate burn severity, we used a delta Normalized Burn Ratio (dNBR; Crocke et al. 2005) cal- culated from downloaded pre-fire (11 July 2014) and post-fire (28 June 2015) images from EarthExplorer Landsat 8 OLI. We clipped and enhanced both images with a linear enhancement using PCI Geomatica software (PCI, Geomatics version 10.1, Richmond Hill, Ontario, Canada). For each image, we calculated a Normalized Burn Ratio (NBR) for each 1 ha raster pixel using the equation: NBR = (NIR (5) – SWIR (7)) / (NIR (5) + SWIR (7)) (Eq. 2) Table 1. Summary of moose GPS locations from 2018 to 2019 that overlapped the Mt. McAllister burn. Average percentages are weighted by the home range area within the burn. HRTot represents the total area in ha of moose home range, HRBurn is the total area in ha of moose home range that overlaps the burn perimeter, HRInBurn is the percentage of moose home range in the burn, PtsTot represents the number of GPS locations for each moose, PtsBurn are the GPS locations for each moose within the burn perimeter, and PtsInBurn is the percentage of GPS locations for each moose within the burn. Animal HRTot (ha) HRBurn (ha) HRInBurn (%) PtsTot PtsBurn PtsInBurn (%) 14,248 1,228 1,201 98 270 270 100 14,255 6,343 2,531 40 294 128 44 14,262 1,160 732 63 263 128 49 14,269 2,112 843 40 206 70 34 14,743 11,646 4,328 37 194 44 23 14,745 232 81 35 257 76 30 14,755 1,208 914 76 250 158 63 Total 23,929 10,630 1,734 874 Average 3,418 1,519 50 248 125 43 SD 4,141 1,445 24 36 75 26 ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 7 where NIR is the near infrared band 5 and SWIR is the shortwave infrared band 7 (Crocke et al. 2005). We used QGIS soft- ware (QGIS Development Team, version 3.1, Open Source Geospatial Foundation Project) to determine the difference between attributes of the 2 images with the equation: dNBR = NBRpre – NBRpost (Eq. 3) where NBRpre is the reflectance values from the pre-fire image and NBRpost is the reflectance values of the post-fire image (Fig. 2). To estimate potential soil moisture, we used a SAGA (System for Automated Geoscientific Analyses) wetness index based on a modified catchment area calcu- lation in ArcMap (Böhner and Selige 2002) to generate soil moisture values for 1 ha raster pixels (Fig. 2). We classified both dNBR (low, medium, and high fire sever- ity) and SAGA (hydric, mesic and xeric wetness) results by calibrating the modelled values found at each site-level field plot to our field classifications from the same plot. To apply the calibrations, we iteratively applied standard numerical classification methods (ESRI 2006) choosing the method that resulted in the most accurate correla- tion with our field classifications; accuracy was the number of agreements divided by total samples (Fawcett 2006). In the case of dNBR values, 72% of the cases agreed with our field classifications using cutpoints that were equidistant across the range of observed values (0.21 and 0.43). In the case of SAGA values, 62% of the cases agreed with our field classification, again using equidistant cutpoints (9 and 12). We accepted that accuracy as adequate based on consistency with standards used in other land classification systems in BC (DeLong et al. 2010). DATA ANALYSIS Site level Pellet group density and the percentage of plants browsed were not distributed nor- mally and could not be transformed to meet the assumptions of normality. Therefore, to test if soil moisture and fire severity influ- enced pellet group density and the percentage of plants browsed, we ranked the data and used a multivariate, two-factor analysis of variance; a multivariate extension of the non- parametric Friedman’s test (Mottonen et al. 2003). We tested collinearity between the two dependent variables using Pearson’s Partial Correlation Coefficient (Zar 2010) and although significant, there was only a weak linear relationship as indicated by the line slope (r = 0.2, P = 0.009). A post-hoc contrast of the main effects on dependent variables was conducted using the Least Significant Difference test (P = 0.05; Zar 2010). Significance of statistical tests were assessed at an alpha of 0.05 and all analyses were con- ducted using SAS version 9.1 2003 (SAS Institute Inc., Cary, North Carolina, USA). Landscape level To test if fire severity and soil moisture influenced use of the burned habitats by each radio-collared moose, we used logistic regression with a logit link function to fit the binomial result (use or random) to fire sever- ity and soil moisture as main effects. Reference classes for soil moisture and fire severity were set as xeric and high, respec- tively. Selection coefficients were averaged across individuals and weighted by the inverse square of the standard error, to give more weight to individuals with more pre- cise data (Dickie et al. 2017, DeMars et al. 2019); this was repeated for the 3 orders of habitat selection. Since availability was not independent of use points in the fourth, daily FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 8 order of selection, we used a conditional logistic regression (Johnson et al. 2006) for that analysis. In logistic regression, selection occurs when the function coefficient (β) is significantly different (P < 0.05) from zero. A positive selection (or preference), occurs when the β is significantly greater than zero (95% CI does not overlap 0) and is the like- lihood of an animal selecting a given item (fire severity and soil moisture) when offered alternative choices (Beyer et al. 2010). Negative selection (avoidance) occurs when β is significantly less than zero. RESULTS Site Level We surveyed a total of 167 vegetation plots and 835 pellet plots on 29 June – 3 July 2018 and 15–24 June 2019. The high- est proportion of plots (36%; 56 of 157) was distributed in areas of low fire severity and mesic soil moisture; the lowest propor- tion (2%) was in areas of high fire severity and xeric soil moisture (Fig. 3). A total of 251 pellet groups were counted with their locations proportional to plot distribution across the classes of fire severity and soil moisture except in two situations: 1) at sites with high fire severity and xeric soil mois- ture where the proportion of pellet groups was low compared to the proportion of plots; and 2) at sites with low fire severity and hydric soil moisture where the propor- tion was high compared to the proportion of plots (Fig. 3). Overall, the proportion of available for- age for moose did not appear to vary across classes of fire severity and soil moisture, with mean forage availability ranging from 60 to 80% (Fig. 4). A total of 44 species were identified with willow having the highest proportion of plants browsed. The propor- tional distribution of plants browsed was less similar to the availability of plots than the proportional distribution of pellet groups, but illustrated a similar trend – lower Fig. 3. The proportion of total sample plots (availability – white bars), moose pellet groups (use – filled dark bars), and percent of plants browsed (use – filled grey bars) within combined classes of fire severity (low, medium, high) and soil moisture (hydric, mesic, xeric). ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 9 proportions in high fire severity and mesic and xeric soil moisture and higher propor- tions in low fire severity and hydric and mesic soil moisture. Pellet group density and the percent of plants browsed varied significantly with fire severity (F4, 302 = 4.13, P = 0.003); but not with soil moisture (F4, 302 = 2.14, P = 0.076). More pellet groups were found at low and medium than high fire severity sites (P < 0.05; Fig. 5A); no difference was found between sites of low and medium fire severity (P > 0.05). Despite the lack of sig- nificance within the multivariate model, pellet group density varied significantly among the soil moisture classifications (P < 0.05; Fig. 5B). Consistent with pellet group density, there was no difference (P > 0.05) between percentage of plants browsed on sites with Fig. 4. The mean proportion and 95% confidence intervals of available moose forage (squares) in (A) fire severity classes (Low, Medium, High) and (B) soil moisture classes (Hydric, Mesic, Xeric) within the Mt. McAllister Burn, 2018–2019. FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 10 low and medium fire severity, but lower browsing occurred on high fire severity sites (P < 0.05; Fig. 6A). Similarly, there was no significant difference between the percent- age of plants browsed on hydric and mesic sites (P > 0.05), but a significant difference between hydric and mesic sites and xeric sites (P < 0.05), (Fig. 6B). Likewise, the percentage of plants browsed on hydric and mesic sites was similar (P > 0.05), whereas a lower percentage (P < 0.05) was found on xeric sites (Fig. 6B). Landscape Level A total of 874 locations from the 7 moose were identified within the burn accounting Fig. 5. The mean pellet group density (pellets/m2, squares) and 95% confidence intervals in (A) fire severity classes (Low, Medium, High) and (B) soil moisture classes (Hydric, Mesic, Xeric) within the Mt. McAllister burn, 2018–2019. ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 11 for 10,630 ha overlap of a total of 23,949 ha encompassing their entire home ranges. Home range size varied by nearly 2 orders of magnitude (~200–12,000 ha), averaging ~3,500 ha. The average overlap was 50% (range = 37–98%; Table 1). At the burn order of selection, moose used areas of low fire severity-mesic/hydric soil moisture more than it was available (Fig. 7A); use and availability were proportional in areas of medium fire severity-mesic soil moisture (Fig. 7A). At the home range order of selection, use and availability were pro- portional, except in the low fire severity- hydric soil moisture areas where use was lower than availability (Fig. 7B). The pro- portion of use and available locations was similar in low/medium fire severity and mesic soil moisture classes, but higher than all other main effect classifications (Fig. 7B); a Fig. 6. The mean percent and 95% confidence intervals of plants browsed in (A) fire severity classes (Low, Medium, High) and (B) soil moisture classes (Hydric, Mesic, Xeric) within the Mt. McAllister burn, 2018–2019. FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 12 similar trend was found at the daily order of selection (Fig. 7C). All areas of high fire severity had proportionally less use than available locations. Although selection models based on the main effects fit the data (P < 0.05), model R2 was low (≤ 0.20; Appendix A). The selection models did not converge for one moose at the burn and home range orders (second and third order) due to its small home range within the burn. Fire severity and soil mois- ture were significant in all models at the Fig. 7. Box (first and third quartiles with median as a line) and whisker (minimum and maximum) plots for the proportion of use (filled) and available (open) moose location classes of fire severity and soil moisture at three levels: (A) daily movements, (B) within collared moose home ranges, and (C) home ranges within the Mt. McAllister burn, 2018–2019. Note that calculation of available within the burn (C) is a total area. ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 13 burn order (second order). At the daily order (fourth order), moisture was significant in 2 models and burn was significant in all (Appendix A). Compared to the reference class of high fire severity, sites with low fire severity were selected for at the burn and home range orders, but neither for nor against at the daily order; moose avoided areas classified as high fire severity at all orders (Fig. 8B, Table 2). Compared to the reference class of xeric soil moisture, sites with hydric soil moisture were selected against at the home range and daily orders, but selected for at the burn order (Fig. 8B, Table 2). At the burn order, mesic soil moisture was neither selected for nor rather than or against, but was selected at the home range and daily orders (Fig. 8B, Table 2). Moose were equivocal in their selection of xeric sites at the home range and daily orders, but avoided these sites at the burn order (Fig. 8B, Table 2). DISCUSSION Site Level Relative to pellet groups and the percent of plants browsed, moose used areas of low to medium fire severity more than areas burned at high severity. Higher use of low and medium fire severity habitats is largely due to patterns of post-fire vegetation succes- sion (Kielland and Brown 2015), and these sites are generally characterized with higher browse availability than unburned sites (MacCracken and Viereck 1990). Although others detected a relationship between post- burn browse availability and relative fire severity (MacCracken and Vierek 1990, Bailey and Whitham 2002), we did not find that relationship. We recommend that other studies examining these relationships include measurements of unburned sites. Moose also used areas that were gener- ally wetter based on measurements of pellet group density and percentage of plants browsed. Although wetter sites may not have proportionally more forage, they mostly con- tained willows and aspens that are preferred forage of moose (Meidinger and Pojar 1991) and may have strongly influenced habitat use. Although not correlated, it appeared that wet- ter sites were burned less severely and would have likely incurred less damage to above- (stems and shoots) and below- (roots) ground plant parts. Habitats with a combination of low to medium fire severity and wetter soil mois- ture tend to provide moose cover as well as forage. Joly et al. (2016) found that the den- sity and basal area of standing live trees were greater on sites of low burn severity than drier and/or more severely burned sites. Therefore, cover would likely be less com- mon on sites with xeric soil moisture and high fire severity, especially in the relatively young McAllister burn. In Alaska, northwest of our study area, greater use of high fire severity sites was documented on older burns during summer months when forage biomass increases and provides cover for moose (Kielland and Brown 2015). There is a general consensus that moose prefer older burns – >11 years old – and that forage pro- duction is highest in burns between 11 and 30 years old (Kelsall et al. 1977, Maier et al. 2005, Joly et al. 2016, Julianus 2016). It is not surprising that we found little use of high fire severity areas given the young age of the McAllister burn, which we expect to change as secondary succession progresses. Landscape Level Habitat use and selection by moose varied by fire severity class with a decreasing trend in selection for low and medium fire severity at a finer resolution of selection (i.e., from home range to daily movements). We also found that fire severity, more than soil mois- ture, influenced where home ranges were and where moose were within their home ranges. FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 14 Fig. 8. Box and whisker plots (first and third quartiles (box) with median (line) and minimum and maximum (whiskers)) for selection coefficients showing the relative preference (values > 0) or avoidance (values < 0) by radio-collared moose for (A) fire severity classes (Low, Medium, High) and (B) soil moisture classes (Hydric, Mesic, Xeric) at three levels: daily movements (filled), within collared moose home ranges (hashed), and home ranges within the Mt. McAllister burn (open), 2018–2019. ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 15 Although we cannot infer that these sta- tistical relationships result in behavioural responses by moose, we did find that all home ranges were located in areas of low fire severity, and moose selectively foraged in areas of low fire severity. At the daily order of selection, they avoided areas of low fire severity suggesting that at a finer selec- tion order, moose used a variety of habitats to fulfill different life requisites. Joly et al. (2016) found that female moose preferred burned habitats and areas with higher solar radiation; more specifically, that maternal females selectively used more forested than riparian areas to minimize predation risk. Moose in the burn ranged in areas that were wetter and avoided more xeric condi- tions. It is not uncommon to find moose in more hydric habitats, as wetlands and lakes are important habitat features providing important life requisites including forage abundance and thermal cover (Wall et al. 2010). When comparing the 3 orders of selection, an increasing trend in use of mesic sites was noted when moving to a finer order (from 2nd to 4th order). At the daily move- ment order, moose preferred mesic more than hydric soil moisture, and showed no selection for xeric sites. Moose use a variety of habitats to fulfill various requirements for growth, reproduc- tion, and survival. Daily use within their home ranges was influenced by burn sever- ity and soil moisture, but use is also influ- enced by multiple factors including sex, pregnancy status, season, winter severity, and other biotic and abiotic factors (Kielland and Brown 2015, Joly et al. 2016). We rec- ognize that our results were biased to winter somewhat and by the single daily location at 0900 hr, and that broader and more frequent locations would likely expand and elucidate patterns of burn use undetected in our study. We found an inconsistency at the daily movement order between the site and land- scape orders. At the site level, moose showed higher use of hydric sites but selected against hydric sites at the daily movement order. In Table 2. Moose selection coefficients and 95% lower and upper confidence intervals (LCI < > average < > UCI) for classes of fire severity (Low, Moderate, High) and soil moisture (Hydric, Mesic, Xeric) at three different orders of selection (Burn, Home range, Daily). Individual radio-collared moose were modelled separately using logistic regression and then averaged for population-level estimates. The number of individual moose was Burn = 6, Home range = 5, and Daily = 6. Selection for or against a factor was defined as confidence intervals that did not overlap zero. Factor Selection order Factor classes Low Medium High Fire severity Burn 0.446 < > 0.684 < > 0.921 0.159 < > 0.349 < > 0.539 −1.346 < > −1.033 < > −0.719 Home range 0.267 < > 0.586 < > 0.906 −0.052 < > 0.16 < > 0.372 −0.955 < > −0.746 < > −0.538 Daily −0.149 < > 0.154 < > 0.457 0.013 < > 0.164 < > 0.315 −0.561 < > −0.318 < > −0.075 Hydric Mesic Xeric Soil moisture Burn 0.058 < > 0.204 < > 0.35 −0.028 < > 0.028 < > 0.083 −0.407 < > −0.231 < > −0.056 Home range −0.498 < > −0.357 < > −0.216 0.157 < > 0.225 < > 0.292 −0.037 < > 0.133 < > 0.302 Daily −0.413 < > −0.258 < > −0.102 0.14 < > 0.19 < > 0.24 −0.092 < > 0.068 < > 0.227 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 16 addition, moose also used areas of low fire severity at the site level, yet showed no pref- erence for low fire severity habitats at the daily movement order. Possible explanations for these inconsistencies are that we had a low sample size of moose at the landscape level and were presumably sampling most moose at the site level. Again, data sets including higher location fix rates may help to identify finer moose movements and hab- itat use patterns (Beyer et al. 2010) that are known to change throughout the day (Poole and Stuart-Smith 2006). In addition, habitat selection is manifested in individual home range size which reflects the spatial avail- ability of resources and ultimately the selec- tion patterns we observed (Herfindal et al. 2009). It is possible that moose with smaller home ranges were using portions of the burn which offered better resources, emphasizing why moose behavior and habitat use should be examined at multiple levels of a spatial scale (Schaefer and Messier 1995, Potochnik and McGill 2012). Conclusions and Management Implications In conclusion, we determined that fire severity, more than soil moisture, influenced habitat use by moose in the recent Mt. McCallister burn. Yet, that influence appeared scale dependent as moose used areas of low fire severity at the course levels of investiga- tion and areas of medium fire severity at the fine levels of investigation, similar to previous research (Mansson et al. 2007). It could be that moose use areas of low fire severity for abundant forage and cover, and more moderately burned (medium fire severity) areas to access forage that provides increased or specific nutritional benefits (Lautenschlager et al. 1997). We suggest that prescribed fires with low and medium fire severity in areas with potential for abundant forage and forest cover would increase habitat value for moose. Such an approach could help restore habitat elements important for moose populations and could be reintro- duced into routine forest management where feasible (Gorley 2016). Our results indicate that fire severity and soil moisture influence moose use of burns and provide a framework for improving prescribed fire monitoring protocols aimed specifically at wildlife habi- tat enhancement (Scasta et al. 2018). Wildfire is a natural landscape distur- bance with long-term influence on moose habitat and population dynamics. Because wildfires are becoming larger and more unpredictable due to climate change and past forest management practices (BCWS 2018), it is critical to recognize related conse- quences for moose. Wildfire management plans, prescribed burning, and strategic fire suppression efforts should consider the impact of fire severity and soil moisture on short- and long-term influences of moose habitat when possible (Gorley 2016). ACKNOWLEDGEMENTS We would like to thank M. Erickson, M. Mueller, and D. Johnston for their help col- lecting data. We would like to thank V. Brumovsky from Wildlife Infometrics for helping with the various GIS work that was required throughout the project. Location data for radio-collared moose was provided through the Peace Region, Fish and Wildlife Compensation Program. Funding for this project was provided by Wildlife Infometrics Inc. We thank L. Husband, P. Pekins, K. Poole, and an anonymous reviewer for com- ments on an earlier draft of this manuscript. REFERENCES Bailey, J. K., and T. G. WhiTham. 2002. Interactions among fire, aspen, and elk affect insect diversity: reversal of a community response. Ecology 83: 1701–1712. ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 17 BCWS. c2018. (accessed November 2018). Beyer, h., D. hayDon, J. morales, J. Frair, m. heBBleWhiTe, m. miTchell, and J. maTThiopoulos. 2010. The interpreta- tion of habitat preference metrics under use-availability designs. Philosophical Transactions of the Royal Society B 365: 2245–2254. Böhner, J., and T. seliGe. 2002. Spatial pre- diction of soil attributes using terrain analysis and climate regionalization. Gottinger Geograpihsche Abhandlungen 115: 13–27. crocKe, a. e., p. Z. Fule, and J. e. crouse. 2005. Comparison of burn severity assessments using differenced normal- ized burn ratio and ground data. International Journal of Wildland Fire 14: 189–198. Dixon, m., h. moBerly, D. WaDe, J. lunDsForD, T. WalDrop, s. GooDricK, c. harper, and G. ToWne. 2018. Introduction to Prescribed Fire in Southern Ecosystems. SRS-054. United States Department of Agriculture, Asheville, North Carolina, USA. (accessed November 2018). DelonG, s. c., h. GriesBauer, W. macKenZie, and V. FoorD. 2010. Corroboration of biogeoclimatic ecosystem classification climate zonation by spatially modelled climate data. British Columbia Journal of Ecosystems and Management 10: 49–64. www.forrex.org/publications/ jem/ISS52/vol10_no3_art7.pdf Demars, c. a., r. serrouya, m. mumma, m. GillinGham, r. mcnay, and s. BouTin. 2019. Moose, caribou and fire: have we got it right yet? Canadian Journal of Zoology. 97: 866–879. DicKie, m., r. serrouya, r. s. mcnay, and s. BouTin. 2017. Faster and farther: wolf movements on linear features and implications for hunting behav- ior. Journal of Applied Ecology 53: 253–263. ESRI. 2006. ArcGIS 9.1 Help topic “stan- dard classification schemes” – Data classification methods. (accessed November 2018). _____. 2015. ArcGIS Desktop: Release 10. Environmental Systems Research Institute, Redlands, California, USA. FaWceTT, T. 2006. An introduction to ROC analysis. Pattern Recognition Letters 27: 861–874. GillinGham, m., and K. parKer. 2008. Differential habitat selection by moose and elk in the Besa-Prophet area of Northern British Columbia. Alces 44: 41–63. GoDDarD, a. 2003. Ungulate winter range species accounts specific to the Peace River region. Woods Environmental Consulting. Fort St. John, British Columbia, Canada. _____. 2011. Peace-Liard Burn Program: Five Year Burn Plan 2012–2017. Fish and Wildlife Branch, Ministry of Forest, Lands, Natural Resource Operations, and Rural Development, Fort St. John, British Columbia, Canada. Gorley, r. 2016. A Strategy to Help Restore Moose Populations in British Columbia. Fish and Wildlife Branch, Ministry of Forest, Lands, Natural Resource Operations, and Rural Development, Victoria, British Columbia, Canada. haTler, D., D. W. noaGorsen, and a. m. Beal. 2018. Carnivores of British Columbia. Royal BC Museum hand- book. The Mammals of British Columbia, Vol. 5. University British Columbia Press, Vancouver, British Columbia, Canada. healy, c., p. J. peKins, l. KanTar, r. conGalTon, and s. aTallah. 2018. https://www2.gov.bc.ca/gov/content/safety/wildfire-status/about-bcws/wildfire-history/wildfire-season-summary https://www2.gov.bc.ca/gov/content/safety/wildfire-status/about-bcws/wildfire-history/wildfire-season-summary https://www2.gov.bc.ca/gov/content/safety/wildfire-status/about-bcws/wildfire-history/wildfire-season-summary https://www2.gov.bc.ca/gov/content/safety/wildfire-status/about-bcws/wildfire-history/wildfire-season-summary https://www.srs.fs.usda.gov/pubs/su/su_srs054.pdf https://www.srs.fs.usda.gov/pubs/su/su_srs054.pdf http://www.forrex.org/publications/jem/ISS52/vol10_no3_art7.pdf http://www.forrex.org/publications/jem/ISS52/vol10_no3_art7.pdf http://webhelp.esri.com/arcgisdesktop/9.1/index.cfm?TopicName=Standard%20classification%20schemes http://webhelp.esri.com/arcgisdesktop/9.1/index.cfm?TopicName=Standard%20classification%20schemes http://webhelp.esri.com/arcgisdesktop/9.1/index.cfm?TopicName=Standard%20classification%20schemes http://webhelp.esri.com/arcgisdesktop/9.1/index.cfm?TopicName=Standard%20classification%20schemes FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 18 Selective habitat use by moose during critical periods in the winter tick life cycle. Alces 54: 85–100. hearD, D. c., D. Walsh, and J. courTier. 2013. Moose Movements in Relation to Roads in Northern Omineca Region, British Columbia. Fish and Wildlife Branch, Ministry of Forest, Lands, Natural Resource Operations, and Rural Development, Fort St. John, British Columbia, Canada. herFinDal, i., J. p. TremBlay, B. BraGe, e. J. solBerF, m. heim, and B. e. saeTher. 2009. Scale dependency and functional response in moose habitat selection. Ecography 32: 849–859. Johnson, a., and p. hale. 2002. The historical foundations of prescribed burning for wildlife: a southeastern perspective. Pages 11–23 in W. M. Ford, K. R. Russell, and C. E. Moorman, editors. Proceedings: The Role of Fire for Nongame Wildlife Management and Community Restoration: Traditional Uses and New Directions. General Technical Report NE-288. United States Department of Agriculture, Newtown Square, Pennsylvania, USA. (accessed November 2018). Johnson, c. J., s. e. nielsen, e. h. merrill, T. l. mcDonalD, and m. s. Boyce. 2006. Resource selection functions based on use-availability data: theoreti- cal motivation and evaluation methods. Journal of Wildlife Management 70: 347–357. Johnson, D. h. 1980. The comparison of usage and availability measurement for evaluating resource preference. Ecology 61: 65–71. Joly, K., m. sorum, T. craiG, and e. Julianus. 2016. The effects of sex, ter- rain and wildfire, winter severity, and maternal status on habitat selection by moose in north-central Alaska. Alces 52: 101–115. Julianus, e. 2016. Moose (Alces alces) browse availability and use in response to post-fire succession on Kanuti National Wildlife Refuge, Alaska. Geography. (accessed November 2018) Kelsall, J. p., e. s. TelFer, and T. D. WriGhT. 1977. The effects of fire on the ecology of the Boreal Forest, with particular reference to the Canadian north: a review and selected bibliography. Canadian Wildlife Service Occasional Paper No. 32. Keeley, J. 2009. Fire intensity, fire severity and burn severity: a brief review and suggested usage. International Journal of Wildland Fire. 18: 116–126. Key, c. h., and n. c. Benson. 2003. The com- posite burn index (CBI): field rating of burn severity. Northern Rocky Mountain Science Center, U.S. Department of the Interior, United States Geological Survey, Reston, Virginia, USA. KiellanD, K., and c. BroWn. 2015. Understanding the Effects of Wildfire Severity on Moose Habitat Characteristics and Use in Interior Alaska. 659 Project Number: 14-3-01-44. Joint Fire Science Program, Institute of Arctic Biology, University of Alaska, Fairbanks, Alaska, USA. KuZyK, G., i. haTTer, s. marshall, c. procTer, B. caDsanD, D. lireTTe, h. schinDler, m. BriDGer, p. sTenT, a. WalKer, and m. KlacZeK. 2018a. Moose population dynamics during 20 years of declining harvest in British Columbia. Alces 54: 101–119. ______, s. marshall, c. procTer, h. schinDler, h. schWanTJe, m. GillinGham, D. hoDDer, s. WhiTe, and m. mumma. 2018b. Determining factors affecting moose population change in British Columbia: testing the landscape change hypothesis. Wildlife Working Report No. WR-126. British Columbia Ministry Forest, Lands, and Natural https://www.fs.usda.gov/treesearch/pubs/19091 https://www.fs.usda.gov/treesearch/pubs/19091 https://www.semanticscholar.org/paper/Moose-(Alces-alces)-browse-availability-and-use-in-Julianus/b3e031420e927786b167ab7cea728625cdc469e6 https://www.semanticscholar.org/paper/Moose-(Alces-alces)-browse-availability-and-use-in-Julianus/b3e031420e927786b167ab7cea728625cdc469e6 https://www.semanticscholar.org/paper/Moose-(Alces-alces)-browse-availability-and-use-in-Julianus/b3e031420e927786b167ab7cea728625cdc469e6 https://www.semanticscholar.org/paper/Moose-(Alces-alces)-browse-availability-and-use-in-Julianus/b3e031420e927786b167ab7cea728625cdc469e6 https://www.semanticscholar.org/paper/Moose-(Alces-alces)-browse-availability-and-use-in-Julianus/b3e031420e927786b167ab7cea728625cdc469e6 ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 19 Resource Operations. Victoria, British Columbia, Canada. lauTenschlaGer, r., h. craWForD, m. sToKes, and T. sTone. 1997. Forest disturbance type differentially affects seasonal moose forage. Alces 33: 49–73. lireTTe, D. 2013. 2013 Winter Moose Survey: MU 7-31. Ministry of Forests, Lands, Natural Resource Operations and Rural Development, Fort St. John, British Columbia, Canada. maccracKen, J., and l. ViereK. 1990. Browse regrowth and use by moose after fire in interior Alaska. Northwest Science 64: 11–18. maier, J., J. m. Ver hoeF, a. D. mcGuire, r. T. BoWyer, l. sapersTein, and h. a. maier. 2005. Distribution and den- sity of moose in relation to land- scape characteristics: effects of scale. Canadian Journal of Forest Research 35: 2233–2243. mansson, J., h. Andren, a. pehrson, and r. BerGsTrom. 2007. Moose browsing and forage availability: a scale-dependent relationship. Canadian Journal of Zoology 85: 372–380. _____, _____, and h. sand. 2011. Can pellet counts be used to accurately describe winter habitat selection by moose Alces alces? European Journal of Wildlife Research 57: 1017–1023. meiDinGer, D., and J. poJar. 1991. Ecosystems of British Columbia. Special Report Series 66. Ministry of Forests, Lands, Natural Resource Operations and Rural Development, Victoria, British Columbia, Canada. Ministry of Forest, Lands, Natural Resource Operations, and Rural Development (MFLNRORD). 2018. Moose in BC fact sheet. Province of British Columbia, Victoria, British Columbia, Canada. (accessed November 2018). moTTonen, J., J. husler, and h. oJa. 2003. Multivariate nonparametric tests in a randomized complete block design. Journal of Multivariate Analysis 85: 106–129. poole, K. G., and K. sTuarT-smiTh. 2006. Winter habitat selection of female moose in western interior montane forests. Canadian Journal of Zoology 84: 1823–1832. poTochniK, a., and B. mcGill. 2012. The limitations of hierarchical organization. Philosophy of Science 79: 120–140. QGIS Development Team. Version 3.1. QGIS Geographic Information System. Open Source Geospatial Foundation Project. (accessed November 2018). saeTher, B-e., and r. Andersen. 1990. Resource limitation in a generalist her- bivore, the moose Alces alces: ecologi- cal constraints on behavioral decisions. Canadian Journal of Zoology 68: 993–999. scasTa, J. D., s. e. r. leVerKus, and m. laVellee. 2018. Peace-Liard Prescribed Fire Monitoring Protocol. Ministry of Forests, Lands, Natural Resource Operations and Rural Development, Fort St. John, British Columbia, Canada. schaeFer, J. a., and F. messier. 1995. Habitat selection as a hierarchy: the spa- tial scales of winter foraging by Muskoxen. Ecography 18: 333–344. scheiDeman, m. c. 2019. Use and selec- tion at two spatial scales by female moose (Alces alces) across central British Columbia following a moun- tain pine beetle outbreak. M. S. Thesis. University of Northern British Columbia, Prince George, British Columbia, Canada. seaman, D., J. millspauGh, B. Kernohan, G. BrunDiGe, K. raeDeKe, and r. GiTZen. 1999. Effects of sample size on kernel home range estimates. The Journal of Wildlife Management 63: 739–747. http://www.env.gov.bc.ca/fw/wildlife/management-issues/docs/2018_moose_fact_sheet.pdf http://www.env.gov.bc.ca/fw/wildlife/management-issues/docs/2018_moose_fact_sheet.pdf http://www.env.gov.bc.ca/fw/wildlife/management-issues/docs/2018_moose_fact_sheet.pdf https://qgis.org/en/site/about/index.html https://qgis.org/en/site/about/index.html FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 20 shacKleTon, D. 2013. Hoofed Mammals of British Columbia. Vol. 3. University British Columbia Press, Vancouver, British Columbia, Canada. siTTler, K. l. 2020. Moose limiting factors investigation: annual report 2015–19. Wildlife Infometrics Inc. Report No. 708. Wildlife Infometrics Inc., Mackenzie, British Columbia, Canada. _____, and s. mcnay. 2017. Moose limiting factors investigation: annual progress report 2016–2017. Report No. 574. Wildlife Infometrics Inc., Mackenzie, British Columbia, Canada. _____, and _____. 2018. Moose (Alces americanus) population density and composition around the Clearwater and Moberly River drainages (Wildlife Management Unit 7-31), Northeastern BC. Report No. 623. Wildlife Infometrics Inc., Mackenzie, British Columbia, Canada. Wall, W. B., m. Belisle, and l. a. luKe. 2010. British Columbia’s interior: moose wildlife habitat decision aid. British Columbia Journal of Ecosystems and Management 11: 45–49. Weil, r., and n. BraDy. 2017. The Nature and Properties of Soils. 15th edition. Pearson Education Inc., Upper Saddle River, New Jersey, USA. Zar, J. h. 2010. Biostatistical Analysis. 5th edition. Pearson Education Inc., Upper Saddle River, New Jersey, USA. ALCES VOL. 57, 2021 FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. 21 A pp en di x A . M od el fi tti ng in fo rm at io n1 , a nd re su lti ng eq ua tio ns , f or m od el s o f t he p ro ba bi lit y of u se o f a si te b y in di vi du al ra di o- co lla re d m oo se (A ni m al ) g iv en th e st at e of m ai n ef fe ct s f or so il m oi st ur e (M oi st ) a nd fi re se ve rit y (D nb r) . A ni m al Pr ef er en ce U se R an do m C on A IC LR Pr M oi st Pr D nb r R Sq C Eq ua tio n 14 ,2 48 bu rn 61 1 13 ,1 40 Y 43 95 0. 00 01 0. 00 40 0. 00 01 0. 14 34 0. 76 7 Lo gi t = − 3. 86 − 0 .2 1( C lM oi st 1) − 0 .0 6( C lM oi st 2) + 1. 71 (C lD nb r1 ) + 0. 30 (C lD nb r2 ) 14 ,2 55 bu rn 12 39 13 ,1 40 Y 77 24 0. 00 01 0. 00 11 0. 00 01 0. 11 13 0. 70 8 Lo gi t = − 2. 91 − 0 .1 5( C lM oi st 1) + 0 .1 5( C lM oi st 2) + 1. 13 (C lD nb r1 ) + 0. 42 (C lD nb r2 ) 14 ,2 62 bu rn 36 2 13 ,1 40 Y 32 76 0. 00 01 0. 00 01 0. 00 01 0. 02 32 0. 61 8 Lo gi t = − 3. 86 + 0 .6 5( C lM oi st 1) + 0 .2 2( C lM oi st 2) + 0. 26 (C lD nb r1 ) − 0 .4 1( C lD nb r2 ) 14 ,2 69 bu rn 41 6 13 ,1 40 Y 34 51 0. 00 01 0. 00 01 0. 00 01 0. 08 42 0. 71 8 Lo gi t = − 4. 01 + 0 .6 3( C lM oi st 1) + 0 .0 7( C lM oi st 2) + 0. 38 (C lD nb r1 ) + 1 .0 7( C lD nb r2 ) 14 ,7 43 bu rn 21 86 13 ,1 40 Y 12 ,0 27 0. 00 01 0. 01 01 0. 00 01 0. 06 20 0. 62 9 Lo gi t = − 1. 93 + 0 .1 0( C lM oi st 1) − 0 .0 8( C lM oi st 2) + 0. 37 (C lD nb r1 ) + 0 .5 4( C lD nb r2 ) 14 ,7 45 bu rn 43 13 ,1 40 N 53 3 0. 00 01 0. 66 05 0. 00 08 0. 09 67 0. 75 6 Lo gi t = − 10 .4 2 + 0. 13 (C lM oi st 1) + 0 .2 1( C lM oi st 2) + 5. 46 (C lD nb r1 ) + 3 .9 8( C lD nb r2 ) 14 ,7 55 bu rn 48 2 13 ,1 40 Y 41 31 0. 00 01 0. 01 05 0. 00 01 0. 01 30 0. 58 5 Lo gi t = − 3. 29 + 0 .2 1( C lM oi st 1) − 0 .1 4( C lM oi st 2) + 0. 24 (C lD nb r1 ) + 0 .1 8( C lD nb r2 ) 14 ,2 48 ho m e ra ng e 27 0 61 1 Y 10 41 0. 00 01 0. 10 06 0. 00 01 0. 08 53 0. 61 3 Lo gi t = − 1. 96 − 0 .1 0( C lM oi st 1) + 0 .2 4( C lM oi st 2) + 1. 21 (C lD nb r1 ) − 0 .5 7( C lD nb r2 ) 14 ,2 55 ho m e ra ng e 12 8 12 39 Y 84 9 0. 02 36 0. 08 24 0. 14 23 0. 01 78 0. 56 9 Lo gi t = − 2. 86 − 0 .4 0( C lM oi st 1) − 0 .0 01 (C lM oi st 2) + 0. 68 (C lD nb r1 ) + 0 .5 2( C lD nb r2 ) 14 ,2 62 ho m e ra ng e 12 8 36 2 Y 55 5 0. 00 11 0. 38 14 0. 00 04 0. 05 34 0. 62 2 Lo gi t = − 1. 23 − 0 .0 3( C lM oi st 1) + 0 .3 7( C lM oi st 2) + 0. 07 (C lD nb r1 ) + 0 .5 3( C lD nb r2 ) 14 ,2 69 ho m e ra ng e 70 41 6 Y 41 0 0. 94 39 0. 82 41 0. 86 58 0. 00 28 0. 52 6 Lo gi t = − 1. 88 − 0 .1 0( C lM oi st 1) + 0 .0 9( C lM oi st 2) − 0. 01 (C lD nb r1 ) + 0 .1 2( C lD nb r2 ) 14 ,7 43 ho m e ra ng e 44 21 86 Y 42 9 0. 00 75 0. 06 19 0. 06 43 0. 03 53 0. 64 4 Lo gi t = − 4. 18 − 0 .8 6( C lM oi st 1) + 0 .1 8( C lM oi st 2) − 0. 39 (C lD nb r1 ) + 0 .4 5( C lD nb r2 ) 14 ,7 45 ho m e ra ng e 76 43 N 16 0 0. 21 52 0. 64 0. 41 26 0. 06 51 0. 59 5 Lo gi t = 4 .9 3 − 0. 38 (C lM oi st 1) − 0 .0 02 (C lM oi st 2) − 4. 26 (C lD nb r1 ) − 5 .0 3( C lD nb r2 ) 14 ,7 55 ho m e ra ng e 15 8 48 2 Y 63 3 0. 00 01 0. 00 56 0. 00 01 0. 19 88 0. 73 3 Lo gi t = − 2 .0 3 − 0. 39 (C lM oi st 1) + 0 .4 3( C lM oi st 2) + 1. 36 (C lD nb r1 ) − 0 .1 4( C lD nb r2 ) 14 ,2 48 da ily 27 0 13 50 Y 93 5 0. 00 01 0. 35 31 0. 00 01 0. 06 00 Lo gi t = − 0 .1 8( C lM oi st 1) + 0 .1 0( C lM oi st 2) + 1. 09 (C lD nb r1 ) − 0 .5 1( C lD nb r2 ) FACTORS INFLUENCING BURN USE BY MOOSE – MCNAY ET AL. ALCES VOL. 57, 2021 22 A ni m al Pr ef er en ce U se R an do m C on A IC LR Pr M oi st Pr D nb r R Sq C Eq ua tio n 14 ,2 55 da ily 12 8 64 0 Y 45 8 0. 07 36 0. 07 47 0. 32 15 0. 02 46 Lo gi t = − 0 .4 6( C lM oi st 1) + 0 .0 6( C lM oi st 2) + 0. 45 (C lD nb r1 ) − 0 .5 5( C lD nb r2 ) 14 ,2 62 da ily 12 8 64 0 Y 44 9 0. 00 15 0. 27 55 0. 00 14 0. 05 03 Lo gi t = 0 .0 1( C lM oi st 1) + 0 .3 2( C lM oi st 2) + 0. 26 (C lD nb r1 ) − 0 .3 5( C lD nb r2 ) 14 ,2 69 da ily 70 35 0 Y 24 1 0. 00 12 0. 48 83 0. 00 06 0. 09 35 Lo gi t = 0 .3 1( C lM oi st 1) + 0 .1 0( C lM oi st 2) − 0. 78 (C lD nb r1 ) + 0 .2 8( C lD nb r2 ) 14 ,7 43 da ily 44 22 0 Y 15 0 0. 00 34 0. 12 93 0. 01 61 0. 12 89 Lo gi t = − 0 .8 3( C lM oi st 1) + 0 .2 0( C lM oi st 2) − 0. 77 (C lD nb r1 ) + 0 .3 6( C lD nb r2 ) 14 ,7 45 da ily 76 38 0 Y 27 3 0. 11 34 0. 29 17 0. 07 09 0. 03 61 Lo gi t = − 0. 42 (C lM oi st 1) + 0 .0 3( C lM oi st 2) + 0. 54 (C lD nb r1 ) − 0 .2 8( C lD nb r2 ) 14 ,7 55 da ily 15 8 79 0 Y 55 1 0. 00 01 0. 01 00 0. 00 08 0. 05 42 Lo gi t = − 0 .4 2( C lM oi st 1) + 0 .0 3( C lM oi st 2) + 0. 54 (C lD nb r1 ) − 0 .2 8( C lD nb r2 ) 1 – W he re : C on is w he th er th e lik el ih oo d es tim at io n co nv er ge d (Y ) o r n ot (N ), A IC is A ka ik e In fo rm at io n C rit er io n, L R is th e lik el ih oo d ra tio , P r M oi st is p ro ba bi lit y of a g re at er C hi -s qu ar e fo r t he m ai n ef fe ct o f s oi l m oi st ur e, P r D nb r i s t he p ro ba bi lit y of a g re at er C hi -s qu ar e fo r t he m ai n ef fe ct o f f ire se ve rit y, R Sq is th e R 2 m od el va lu e, C is th e ar ea u nd er th e re ce iv er o pe ra tin g ch ar ac te ris tic c ur ve . T he p ro ba bi lit y of a si te b ei ng u se d gi ve n th e st at e of m ai n ef fe ct s i s 1 / (1 + e lo gi t ) w he re th e lo gi t va lu e is fo un d by e ac h m od el e qu at io n.