FINE-SCALE TEMPERATURE PATTERNS IN THE SOUTHERN BOREAL FOREST: IMPLICATIONS FOR THE COLD-ADAPTED MOOSE Bryce Olson1,2, Steve K. Windels1, Mark Fulton2, and Ron Moen3 1National Park Service, Voyageurs National Park, 360 Highway 11 E, International Falls, Minnesota 56649; 2Bemidji State University, 1500 Birchmont Drive NE, Bemidji, Minnesota 56601; 3Natural Resources Research Institute, University of Minnesota, 5013 Miller Trunk Highway, Duluth, Minnesota 55811 ABSTRACT: Moose (Alces alces) respond to warm temperatures through both physiological and behavioral mechanisms. Moose can reduce heat load via habitat selection when spatial and temporal variation exists within the thermal environment. We recorded operative temperatures (To) throughout the Kabetogama Peninsula of Voyageurs National Park, Minnesota for 1 year to describe seasonal pat- terns in the thermal environment available to moose and identify physical and landscape characteristics that affect To in southern boreal forests. Significant predictors of To varied by season and time of day and included vegetation cover type, canopy cover, and slope/aspect. Vegetation cover type influenced To during summer and fall afternoons with additional variation during summer afternoons explained by percent canopy cover. Slope/aspect was the main driver of To during winter and spring afternoons. Slope position was not a significant predictor of temperature, likely because of low topographic relief in our study area. The Tos were significantly warmer in open versus closed habitats during the day with the pattern reversed at night. Our results can be used to test if moose display a behavioral response to To at various spatial and temporal scales. ALCES VOL. 50: 105–120 (2014) Key words: Alces alces, aspect, canopy cover, cover type, forests, moose, operative temperature, topography. Moose (Alces alces) are adapted to cold environments (Karns 2007) but conversely, are less tolerant of high ambient tempera- tures (Ta). Renecker and Hudson (1986) esti- mated the upper critical temperatures (Tuc) of moose as −5 °C in winter and 14 °C in sum- mer, with open mouth panting occurring at 0 °C and 20 °C, respectively. A recent study estimated a slightly higher Tuc in summer (17 °C; McCann et al. 2013). These esti- mates of Tuc provide a lower limit of Ta at which moose presumably employ physiolo- gical and behavioral mechanisms to reduce thermal stress. Moose respond physiologically to high Ta by reducing metabolic rate, flattening their pelage, and increasing respiratory rate to expel excess heat, but they cannot sweat (Schwartz and Renecker 2007). They also exhibit behavioral responses including higher use of conifer stands for thermal refuge and nocturnal activity (DeMarchi and Bunnel 1995, Dussault et al. 2004, Broders et al. 2012). Chronic exposure of moose to high Ta has been correlated with reduced weight gain in Norway (van Beest and Milner 2013), lower survival in northeastern Minne- sota (Lenarz et al. 2009), population declines in northwestern Minnesota (Murray et al. 2006), and distribution shifts in China (Dou et al. 2013). Fine-scale differences in Ta likely exist across space and time at the southern extent of moose range, and indivi- dual moose should exploit these differences to mitigate the effects of high Ta on body 105 condition and ultimately fitness. Previous stu- dies of moose habitat selection and Ta focused on forest cover type as the main driver of ther- mal conditions across the landscape (e.g., Lowe et al. 2010). Other factors affecting variability in the thermal landscape include elevation, canopy cover, slope and aspect, and position on slope (Reifsnyder et al. 1971, Chen and Franklin 1997, Danielson et al. 1997, Chen et al. 1999, Ellis and Pomeroy 2007). Habitat selection patterns relative to Ta cannot be fully understood with- out a clear understanding of patterns in the thermal environment at different spatial and temporal scales. Operative temperature (To) is an approx- imation of the convective and radiant heat transfer on the surface of an animal, making it a useful measure to interpret the thermal environment experienced by animals versus Ta alone (Dzialowski 2005). For example, animals experience different To in sunlight, wind, or under forest canopy at the same Ta. It is easiest to estimate To with a black globe thermometer (Vernon 1930, 1932, 1933; Fig. 1) which consists of a matte black painted copper sphere containing a tempera- ture logger that integrates Ta, mean radiant temperature, and air movements into a single metric (Bedford and Warner 1934). Our objectives were to identify physical and vegetative factors that influence To, and to characterize the thermal environment experienced by moose across different cover types in Voyageurs National Park (VNP) in northeastern Minnesota. STUDY AREA Voyageurs National Park (VNP) is situ- ated on the southern limit of North American Fig. 1. Cross-section and attached black globe thermometer showing Hobo U22 Water Temp Pro v2 temperature loggers inserted into a 15 cm diameter copper toilet bowl float painted matte black. Loggers were hung 0.75 m above ground and 15 cm from the trunk on the northeast side of a tree. 106 FOREST TEMPERATURE PATTERNS – OLSON ET AL. ALCES VOL. 50, 2014 moose range, along the Minnesota-Ontario border (Fig. 2). The climate is mid-continental with long cold winters and short cool sum- mers. Mean monthly temperatures range from −15 °C in January to 19 °C in July with an annual mean temperature of 3 °C (NOAA 2010). First snowfall usually occurs in early November and final snow- fall in early April. Average annual precipi- tation is 61 cm, with an average annual snowfall of 183 cm (NOAA 2010). We limited the study area to the 329 km2 Kabetogama Peninsula as this is where most moose in VNP currently reside (Windels 2014, Fig. 2). Vegetation in the Kabetogama Peninsula is typical of the southern boreal and Laurentian mixed conifer-hardwood regions (Faber-Langendoen et al. 2007). For- est cover is a mosaic of quaking aspen (Populus tremuloides), paper birch (Betula papyrifera), balsam fir (Abies balsamea), and jack (Pinus banksiana), red (P. resi- nosa), and white pine (P. strobus). A variety of wetlands including bogs, fens, marshes, and swamps are interspersed across the land- scape (Faber-Langendoen et al. 2007). Geo- logical features include thin and sandy topsoil with regions of exposed bedrock (Ojakangas and Matsch 1982). Moose and white-tailed deer (Odocoileus virginianus) are the only ungulate species in VNP; woodland caribou (Rangifer tarandus) were extirpated by the early 1900s (Cole 1987). Moose density in the Kabetogama Peninsula ranges from 0.14–0.19 moose/km2 and has remained stable since the 1990s (Windels 2014). Beaver (Castor canadensis) are abundant and contribute significantly to the spatial heterogeneity of the landscape (Johnston and Naiman 1990). A variety of human and other natural dis- turbances have created a diverse mosaic of Fig. 2. Location of the Kabetogama Peninsula study area in Voyageurs National Park, Minnesota, USA. ALCES VOL. 50, 2014 OLSON ET AL. – FOREST TEMPERATURE PATTERNS 107 vegetation in multiple seral stages. Wildfires and extensive logging occurred in the 1920s and 1930s, followed by less intensive log- ging through the 1960s (Gogan et al. 1997). Major fires occurred throughout VNP in 1923 and 1936. Suppression of fire followed until the late 1980s when the National Park Service implemented a wildland fire man- agement plan, though most prescribed burns have been relatively small (National Park Service, unpublished data). METHODS To measure To across the Kabetogama Peninsula, we stratified our sampling design by landscape and vegetation characteristics identified within a 30-m × 30-m pixel matrix that matches with LandSat imagery. We derived slope and aspect for each pixel using 30 m resolution Shuttle Radar Topo- graphy Mission data (version 1) (Rabus 2003, Rodriguez et al. 2005, 2006). We esti- mated slope using the slope function from ERDAS Imagine (ERDAS Inc. 2010). We categorized slope as either <10% (5.7°) or ≥10% (5.7°). We calculated aspect using the aspect function of ERDAS Imagine (ERDAS Inc. 2010) and classified each pixel into 1 of 2 categories: aspects between 315-45° (i.e., north) and aspects between 45-315° (i.e., east/south/west). We assumed that solar inputs would be lower on north facing slopes compared to east, south, and west facing slopes. Therefore, we combined slope and aspect into flat, slopes facing north, and slopes facing east/south/west for analysis. Detectable variation in Ta as a function of elevation was not expected in VNP as most local relief is <30 m and maximum relief within the Peninsula is only 81 m. Therefore, elevation was not considered in our sampling design or subse- quent modeling. We developed a canopy cover model using the methodology outlined in the Great Lakes Inventory and Monitoring Network’s Landscape Dynamics protocol (Kennedy and Kirschbaum 2010, Kennedy et al. 2010). Percent cover of trees, shrub, and ground layer was estimated at 30-m pixels in ArcMap (ESRI 2011) using high-resolution air photos taken in the spring (leaf-off) and summer (leaf-on) of 2008 with 0.15 and 1 m resolu- tions, respectively (Kirschbaum and Gafvert 2010). Estimates of canopy cover in each pixel were related to the normalized burn ratio (van Wagtendonk et al. 2004) calcu- lated from the Landsat image corresponding to that time period to create a regression model of canopy cover. We categorized canopy cover as open (i.e., no or few canopy-forming trees), variable cover (i.e., non-forested, discontinuous canopy), <70% forest cover, 70–80% forest cover, and >80% forest cover. We classified vegetation cover type using the National Vegetation Classification System Subclass Level (decid- uous, evergreen, mixed, woodland, shrub, or herbaceous) developed for VNP (Faber- Langendoen et al. 2007). We developed a sampling matrix using the 3 sets of variables (vegetation cover type, canopy cover, slope/aspect). Each unique combination of the 3 variables was given an identifying code using ERDAS Imagine. A 30-m pixel raster map was cre- ated and converted to a polygon shapefile. Areas for each polygon were calculated and polygons <1.07 ha (12 pixels) were deleted to avoid sampling very small patches. We sampled at an intensity of 1 temperature log- ger for every 333 ha in the study area. We randomly selected polygons to sample from each of 38 unique combinations of vegeta- tion cover type, canopy cover class, and slope/aspect class. We located the sample point near the centroid of the polygon to allow a sufficient buffer between adjacent polygons. We used alternate sites if the selected polygon centroid was <30 m (1 pixel) from the edge, field reconnaissance found that site characteristics were different 108 FOREST TEMPERATURE PATTERNS – OLSON ET AL. ALCES VOL. 50, 2014 from remotely-sensed data, or sites were otherwise inaccessible (e.g., flooding). Black globe temperature loggers con- sisted of a data-logging thermocouple (Onset Computer Corporation, Bourne, Massachu- setts, USA) inserted into a copper toilet tank float painted matte black (Fig. 1). We calibrated loggers for a minimum of 96 h to verify accuracy and resolution as compared to the stated equipment specifications of the logger (±0.21 °C from 0°−50 °C). All log- gers were synchronized and programmed to record temperatures every 15 min for 1 year. At each sample point, we hung loggers 0.75 m above the ground and 15 cm from the trunk. Loggers were placed on the north- east side of trees to minimize direct solar radiation during the warmest time of day (Fig. 1). We used handheld field computers with GPS to verify that logger placement in the field was consistent with identified cover type and location within the cover type poly- gon. We used real-time GIS and measure- ments in the field to ensure we were within the identified cover type, canopy cover class, and slope/aspect category before deploying loggers. Loggers were deployed from June 2010 to July 2011, with periodic downloads to reduce risk of data loss. Data were screened to remove biased or failed measurements (e.g., faulty logger, damaged globe or log- ger, and snow-covered loggers). We deployed an additional set of loggers from August 2011 to January 2012 to test for differences in position on slope. We ran- domly deployed 3 loggers in each of 9 com- binations of cover type (deciduous, evergreen, and mixed), canopy cover class (<70%, 70–80%, >80% forested canopy), and slope position (top, mid-slope, and base); slopes ranged from 17–47%. We analyzed data by season with full factorial repeated-measures ANOVA with type III sums of squares using SPSS 20 (IBM Corporation 2011). We defined sea- sons as spring (1 March–31 May), summer (1 June–31 August), fall (1 September–30 November), and winter (1 December–28 February). Each 15 min logger interval was treated as the response variable but con- trolled for repeated measures. Post-hoc pair- wise comparisons were made using the Bonferroni method. Significance levels were set to P = 0.05 for all tests. Sub- samples of the dataset were made to compare differences during the 3 warmest hours of the day (1300–1600 hr) and the 3 coldest hours of the day (0300–0600 hr). We used a similar statistical approach to test for differences in slope position for summer and winter only. Based on model results, we assessed the availability of potential thermal refugia to moose across the Kabetogama Peninsula. We simulated moose home ranges by creat- ing 25 random points within the study area and then buffering those points to approxi- mate mean annual moose home ranges in VNP (48 km2 ± 33.5 SD) as reported by Cobb et al. (2004). Home ranges varied in size due to the highly variable shoreline of the Kabetogama Peninsula. Within each simulated home range, we estimated the pro- portion of each habitat type (vegetation cover type, canopy cover class, slope/aspect) and used summary statistics to highlight availability of selected habitat features. RESULTS Open habitat types (shrub and herbac- eous) were significantly warmer than forested habitat types during the summer (maximum difference = 3.38 °C; Fig. 3) with the greatest difference occurring in the afternoon (maximum difference = 8.10 °C; Table 1, Fig. 3). Open habitat types were also the coolest at night (maximum differ- ence = 2.66 °C; Table 1, Fig. 3). Mean To during the summer did not differ among forested cover types for any of the time per- iods (Daily, Hot, Cold). Herbaceous cover types were warmer than forested cover types in the afternoon during the fall (maximum ALCES VOL. 50, 2014 OLSON ET AL. – FOREST TEMPERATURE PATTERNS 109 difference = 7.43 °C; Table 1, Fig. 3). Tem- peratures in shrub cover types were inter- mediate between herbaceous and forested cover types. The amount of canopy influ- enced To within forested cover types. Areas with >80% canopy coverage were cooler than those with <70% cover (Table 2, Fig. 4); To in the 70–80% cover class was interme- diate to these. Slope/aspect influenced To only during the 3 coldest hours of day in summer. Flat areas were cooler than East/South/West facing slopes (difference = 1.14 °C; Table 3). North-facing slopes were intermediate between flat areas and East/South/West facing slopes. Slope/aspect had no influence on To during the fall months. Winter temperature was only differentiated by slope/aspect during the afternoon with northern facing slopes cooler than both flat and east/south/west categories (Table 3, Fig. 5). Spring temperatures varied across slope/aspect categories during the afternoon hours. North-facing slopes were cooler than flat areas (Table 3, Fig. 5). To was not different among slope positions in summer (F2,21 = 0.287, P = 0.755) or winter (F2,21 = 0.606, P = 0.556). The majority of the Kabetogama Penin- sula consists of forested cover types with >70 percent canopy cover and flat topogra- phy (Table 4, Fig. 6, 7). Simulated home ranges varied in size from 23–57 km2 with a mean of 46 km2 (SD = 10.1 km2). Potential summer refugia, such as high-canopy cover forests, were found in about 40% of simu- lated home ranges (Table 4). However, north-facing slopes are relatively limited in the study area and the percentage of north- facing slopes in simulated home ranges was <10% (Table 4, Fig. 7). DISCUSSION Vegetation cover type, percent canopy cover, and slope/aspect all influenced To, although differently depending on season and time of day. Vegetation cover type had the strongest influence on To during summer months and fall afternoons. This was largely Fig. 3. Mean operative temperatures across vegetation cover types in summer over a 24-hour period, 1 June–31 August 2010, Voyageurs National Park, Minnesota, USA. 110 FOREST TEMPERATURE PATTERNS – OLSON ET AL. ALCES VOL. 50, 2014 Table 1. Mean daily (24-hour mean) operative temperatures (°C) across vegetation cover types in spring (1 March–31 May 2011), summer (1 June–31 August 2010), fall (1 September–30 November 2010), and winter (1 December 2010–28 February 2011), Voyageurs National Park, Minnesota, USA. Mean operative temperature for the 3 warmest (Hot) and 3 coldest (Cold) hours of the day are also shown. Means followed by the same letter within a row are not significantly different from each other. Vegetation Cover Type Significance Season Deciduous Evergreen Mixed Woodland Shrub Herbaceous F5,85 P Value Sprint Daily 3.97 ± 0.74 ab 2.89 ± 0.78 a 4.28 ± 0.73 ab 3.53 ± 1.81 ab 5.26 ± 1.48 ab 6.70 ± 1.15 b 0.86 0.496 Hot 10.47 ± 0.90 a 7.95 ± 0.95 b 10.29 ± 0.90 ab 9.23 ± 2.21 abc 14.19 ± 1.80 ac 16.63 ± 1.40 c 2.59 0.051 Cold −1.45 ± 0.65 a −1.71 ± 0.68 a −1.25 ± 0.64 a −1.71 ± 1.59 a −1.64 ± 1.01 a −2.39 ± 1.30 a 0.14 0.934 Summer Daily 18.32 ± 0.33 a 18.76 ± 0.35 a 18.42 ± 0.33 a 19.33 ± 0.60 ab 21.25 ± 0.60 bc 21.70 ± 0.42 c 2.93 0.029 Hot 23.03 ± 0.57 a 23.66 ± 0.60 a 23.13 ± 0.57 a 24.52 ± 1.04 a 29.84 ± 1.04 b 31.13 ± 0.73 b 7.10 <0.001 Cold 13.76 ± 0.37 a 13.97 ± 0.39 ab 13.54 ± 0.37 ab 14.68 ± 0.67 b 12.88 ± 0.47 ab 12.02 ± 0.67 a 2.86 0.045 Fall Daily 5.56 ± 0.65 a 5.56 ± 0.69 a 5.66 ± 0.67 a 5.99 ± 1.18 a 6.38 ± 1.18 a 7.49 ± 0.83 a 0.32 0.862 Hot 9.97 ± 0.82 ab 9.26 ± 0.87 ab 9.77 ± 0.85 ab 10.48 ± 1.49 ab 14.24 ± 1.49 ac 16.69 ± 1.05 c 3.00 0.026 Cold 2.48 ± 0.57 a 2.78 ± 0.60 a 2.63 ± 0.59 a 2.93 ± 1.03 a 2.23 ± 0.73 a 1.52 ± 1.03 a 0.35 0.786 Winter Daily −13.37 ± 0.24 a −12.99 ± 0.25 a −13.41 ± 0.25 a −13.25 ± 0.44 a −14.58 ± 0.44 a −14.20 ± 0.36 a 1.70 0.166 Hot −9.20 ± 0.40 ab −9.68 ± 0.43 a −9.72 ± 0.42 a −9.15 ± 0.73 ab −9.08 ± 0.73 ab −6.95 ± 0.60 b 2.14 0.091 Cold −16.17 ± 0.33 a −15.59 ± 0.35 a −16.02 ± 0.34 a −16.10 ± 0.60 a −17.22 ± 0.49 a −17.30 ± 0.60 a 1.19 0.324 A L C E S V O L . 5 0 , 2 0 1 4 O L S O N E T A L . – F O R E S T T E M P E R A T U R E PA T T E R N S 1 1 1 Table 2. Mean daily (24-hour mean) operative temperatures (°C) across varying amounts of canopy cover by season in spring (1 March–31 May 2011), summer (1 June–31 August 2010), fall (1 September–30 November 2010), and winter (1 December 2010–28 February 2011), Voyageurs National Park, Minnesota, USA. Mean operative temperature for the 3 warmest (Hot) and 3 coldest (Cold) hours of the day are also shown. Means followed by the same letter within a row are not significantly different from each other. Canopy Cover Significance Season Open Variable <70% 70–80% >80% F 4,85 P Value Spring Daily 6.70 ± 1.15 a 4.40 ± 1.17 a 4.47 ± 0.75 a 3.14 ± 0.77 a 3.47 ± 0.72 a 0.59 0.561 Hot 16.46 ± 1.35 a 11.67 ± 1.38 a 10.99 ± 0.92 ab 8.66 ± 0.94 bc 9.00 ± 0.88 bc 1.38 0.265 Cold −1.644 ± 1.01 a −2.05 ± 1.03 a −1.27 ± 0.66 a −1.87 ± 0.68 a −1.31 ± 0.64 a 0.25 0.783 Summer Daily 21.70 ± 0.42 a 20.29 ± 0.42 a 19.00 ± 0.35 b 18.54 ± 0.34 bc 17.96 ± 0.31 c 2.48 0.093 Hot 30.93 ± 067 a 27.09 ± 0.67 b 24.39 ± 0.61 c 23.24 ± 0.59 cd 22.18 ± 0.54 d 3.71 0.031 Cold 12.88 ± 0.47 a 13.35 ± 0.47 a 13.49 ± 0.39 a 13.95 ± 0.38 a 13.83 ± 0.35 a 0.38 0.688 Fall Daily 7.48 ± 0.83 a 6.18 ± 0.83 a 5.57 ± 0.69 a 5.73 ± 0.67 a 5.48 ± 0.65 a 0.04 0.965 Hot 16.39 ± 1.01 a 12.36 ± 1.01 a 10.25 ± 0.88 b 9.79 ± 0.85 b 8.95 ± 0.81 b 0.62 0.541 Cold 2.23 ± 0.73 a 2.23 ± 0.73 a 2.40 ± 0.61 a 2.69 ± 059 a 2.81 ± 0.56 a 0.13 0.879 Winter Daily −14.20 ± 0.36 a −13.92 ± 0.31 a −13.45 ± 0.26 a −13.28 ± 0.25 a −13.07 ± 0.24 a 0.57 0.569 Hot −7.11 ± 0.53 a −8.82 ± 0.46 a −8.94 ± 0.43 a −9.70 ± 0.42 b −10.02 ± 0.40 b 1.75 0.184 Cold −17.22 ± 0.49 a −16.70 ± 0.42 a −16.29 ± 0.35 a −15.98 ± 0.34 a −15.50 ± 0.33 a 1.39 0.259 1 1 2 F O R E S T T E M P E R A T U R E PA T T E R N S – O L S O N E T A L . A L C E S V O L . 5 0 , 2 0 1 4 Fig. 4. Mean operative temperatures across varying amounts of canopy cover in summer over a 24-hour period, 1 June–31 August 2010, Voyageurs National Park, Minnesota, USA. Table 3. Mean daily (24-hour mean) operative temperatures (°C) across slope/aspect categories in spring (1 March–31 May 2011), summer (1 June–31 August 2010), fall (1 September–30 November 2010), and winter (1 December 2010–28 February 2011), Voyageurs National Park, Minnesota, USA. Mean operative temperature for the 3 warmest (Hot) and 3 coldest (Cold) hours of the day are also shown. Means followed by the same letter within a row are not significantly different from each other. Slope / Aspect Significance Season Flat East/South/West North F2,85 P Value Spring Daily 4.15 ± 0.39 a 4.48 ± 0.88 a 2.87 ± 0.91 a 0.61 0.551 Hot 11.14 ± 0.48 a 11.05 ± 1.07 ab 7.27 ± 1.10 b 3.36 0.045 Cold −1.79 ± 0.35 a −1.07 ± 077 a −1.57 ± 0.80 a 0.20 0.816 Summer Daily 18.79 ± 0.16 a 19.01 ± 0.37 a 18.35 ± 0.40 a 2.16 0.125 Hot 24.73 ± 0.28 a 24.03 ± 0.65 ab 22.30 ± 0.69 b 1.76 0.181 Cold 13.03 ± 0.18 a 14.17 ± 0.42 b 14.11 ± 0.45 ab 4.92 0.011 Fall Daily 5.32 ± 0.33 a 6.34 ± 0.76 a 5.56 ± 0.79 a 1.54 0.225 Hot 10.60 ± 0.41 a 10.92 ± 0.96 a 8.55 ± 0.99 a 1.50 0.233 Cold 1.86 ± 0.29 a 3.08 ± 0.67 b 3.07 ± 0.69 ab 2.60 0.084 Winter Daily −13.41 ± 0.12 a −12.84 ± 0.28 a −13.74 ± 0.29 a 2.45 0.097 Hot −8.89 ± 0.21 a −8.35 ± 0.47 a −11.22 ± 0.49 b 10.24 <0.001 Cold −16.18 ± 0.17 a −15.84 ± 0.39 a −15.99 ± 0.40 a 0.041 0.959 ALCES VOL. 50, 2014 OLSON ET AL. – FOREST TEMPERATURE PATTERNS 113 Fig. 5. Mean operative temperatures across slope/aspect categories in winter over a 24-hour period, 1 December 2010–28 February 2011, Voyageurs National Park, Minnesota, USA. Table 4. Mean, standard deviation (SD), minimum (Min), and maximum (Max) percentage of vegetation cover type, canopy cover, and slope/aspect categories in 25 simulated moose home ranges in Voyageurs National Park, Minnesota, USA. Simulated Home Range Variable Mean SD Min Max Vegetation Cover Type Evergreen 13 5 6 23 Deciduous 34 11 20 58 Mixed 27 7 14 48 Woodland 4 2 1 7 Shrub 6 4 2 16 Herbaceous 16 8 8 36 Canopy Cover Class High 41 6 29 52 Med 23 6 13 34 Low 10 4 4 15 Variable 10 3 5 16 Open 16 8 8 36 Slope / Aspect East/South/West 8 4 2 13 Flat 87 5 79 96 North 5 2 2 8 114 FOREST TEMPERATURE PATTERNS – OLSON ET AL. ALCES VOL. 50, 2014 driven by open versus forested habitats, although we detected small but significant differences in To within closed forest habi- tats, similar to other studies (e.g., McGraw et al. 2012). Amount of canopy cover signif- icantly affected To only during afternoons in the summer months. Forest type and the amount of canopy cover effectively combine to reduce the amount of solar radiation that reaches the forest floor and therefore can reduce heat loading from direct solar radia- tion (Demarchi and Bunnel 1993). Vegeta- tion volume, hence canopy cover, is greatest in summer months, likewise solar angle is most direct during summer after- noons. Areas with thick vegetation and dense canopy cover may serve as ideal thermal refuge for moose during the day (DeMarchi and Bunnell 1995, Dussault et al. 2004, van Beest et al. 2012). Although some variation exists in the amount of forested habitat types with high canopy cover within our simulated home ranges, these habitat types do not seem limited in the study area. Open cover types were cooler than forested cover types during the 3 coldest hours of the day. Dense vegetation and canopy cover actually retain heat within forested cover types while open cover types release more heat at night (Chen et al. 1993). Moose in central Norway use open habitat types at night and older forested stands during daytime (Bjørneraas et al. Fig. 6. Spatial distribution of potential summer thermal refugia across the Kabetogama Peninsula in Voyageurs National Park, Minnesota, USA. Forested areas with >80% canopy cover are coolest during the day (black pixels) while herbaceous and shrub cover types (gray pixels) are coolest at night. All other habitat types are shown as white. Inset shows fine-scale juxtaposition of “cool” and “hot” habitats at 30-m pixel resolution. ALCES VOL. 50, 2014 OLSON ET AL. – FOREST TEMPERATURE PATTERNS 115 2011). The availability of open cover types may be limited for some moose in the Kabetogama Peninsula. We defined 4 equal seasons in our mod- els based on calendar months rather than the timing of leaf phenology which varies annually in response to weather events, dis- ease, drought, and other factors (Lechowicz 1984). As a consequence, our ability to detect significant differences may have been diminished for some variables, specifi- cally canopy cover. Future studies of To should consider incorporating important pre- dictor variables that may change at relatively fine time scales. Also, canopy cover esti- mates were based on the leaf-on period, and may not accurately reflect the true amount of canopy cover during leaf-off periods for all cover types with a deciduous tree component. The majority of the study area was flat and cooler at night than east/south/west facing slopes during summer, likely due to differences in radiant heat loss. Slope/aspect was the only significant influence on To during winter months as well as spring afternoons. Slope/aspect may have a stronger effect on To during winter months when the solar angle is at its lowest. These environments may serve as thermal refugia on warm days in winter and early spring as topographic exposure can influence maximum daily tem- peratures (Bolstad et al. 1998). Additionally, radiation received on flat and south facing slopes may be reflected to the body by the high reflectivity of snow in winter. More Fig. 7. Spatial distribution of potential winter and spring thermal refugia (i.e., north- facing slopes; black pixels) across the Kabetogama Peninsula in Voyageurs National Park, Minnesota, USA. Areas with flat aspect are shown for comparison (gray pixels). All other habitat types are white. Inset shows fine-scale juxtaposition of north-facing slopes with other aspects at 30-m pixel resolution. 116 FOREST TEMPERATURE PATTERNS – OLSON ET AL. ALCES VOL. 50, 2014 northerly locations may realize increased effect of slope/aspect, as well as areas with greater topographic relief. Moose strongly selected north-facing slopes in southwestern Alberta due to increased shade and browse availability (Telfer 1988). North-facing slopes make up less than 5 percent of our study area and across most simulated home ranges, sug- gesting this type of seasonal thermal refugia may be limited for moose in our study area. We did not detect an effect of slope posi- tion on To, presumably because of the low topographic relief in the study area (Daniel- son et al. 1997). Areas with larger eleva- tional gradients than our study area should include elevation as a variable due to the adiabatic lapse or rate change in temperature of an air mass as it changes with altitude (American Meteorological Society 2000). In certain studies elevation was the single strongest driver of temperature difference (e.g., Lookingbill and Urban 2003). Moose use aquatic habitats for a variety of reasons including foraging, sodium acqui- sition, insect relief, and thermoregulation (Peek 2007). Aquatic habitats in our study area contain little to no canopy cover and related To regimes are likely similar to that of open habitat types. Although moose using shallow, aquatic habitats during daytime may be exposed to direct solar radiation, they could mitigate heat loading by submerging in water. Thermal variability exists at relatively fine scales across our study area due primar- ily to the fine mosaic of vegetation cover types, canopy coverage, and site aspect (Fig. 6, 7). We detected maximum differences in mean To of ≤9 °C across all habitat types during the warmest parts of summer days. Within forested habitat types, there was >2 °C difference across canopy cover cate- gories in summer. Slope/aspect accounted for as much as a 4 °C difference in To during winter and spring. Even small differences in the thermal environment may be relevant for achieving individual heat balance (Renecker and Hudson 1990). The availability of thermal refugia will be of greater importance at the southern edge of moose range as mean annual tem- perature continues to rise with climate change (IPCC 2007). Behavioral responses to high Ta include specific microhabitat use and activity shifts in other parts of moose range (Dussault et al. 2004, Broders et al. 2012, van Beest et al. 2012). To mitigate the effects of increasing Ta, managers should promote a variety of habitat types to provide adequate thermal refugia within a typical home range while meeting other life history requirements of moose (Peek 2007). ACKNOWLEDGEMENTS We thank B. Severud, J. Warmbold, C. Eckman, D. Morris, and N. Walker for assis- tance with field work and A. Kirschbaum for remote sensing support. This project was funded by Voyageurs National Park, a grant from the USGS-NPS Natural Resource Pre- servation Program, a grant from the U.S. National Park Service’s Great Lake Research and Education Center, Bemidji State Univer- sity, the Natural Resources Research Insti- tute at the University of Minnesota Duluth, and the Environment and Natural Resources Trust Fund. REFERENCES AMERICAN METEOROLOGICAL SOCIETY. 2000. Glossary of Meteorology. Second edition. Allen Press, New York, New York, USA. BEDFORD, T., and C. G. WARNER. 1934. The globe thermometer in studies of heating and ventilation. Journal of Hygeine 34: 458. BOLSTAD, P. V., L. SWIFT, F. COLLINS, and J. REGNIERE. 1998. Measured and predicted air temperatures at basin to regional scales in the southern Appalachian mountains. Agricultural and Forest Meteorology 91: 161–176. ALCES VOL. 50, 2014 OLSON ET AL. – FOREST TEMPERATURE PATTERNS 117 BJØRNERAAS, K., E. J. SOLBERG, I. HERFINDAL, B. V. MOORTER, C. M. ROLANDSEN, J. TREMBLAY, C. SKARPE, B. SAETHER, R. ERIKSEN, and R. ASTRUP. 2011. Moose Alces alces habitat use at multiple tem- poral scales in a human-altered land- scape. Wildlife Biology 17: 44–54. BRODERS, H. G., A. B. COOMBS, and J. R. MCCARRON. 2012. Ecothermic responses of moose (Alces alces) to thermoregula- tory stress on mainland Nova Scotia. Alces 48: 53–61. CHEN, J., and J. F. FRANKLIN. 1997. Growing- season microclimate variability within an old-growth Douglas fir forest. Climate Research 8: 21–34. ———, ———, and T. A. SPIES. 1993. Con- trasting microclimates among clearcut, edge, and interior old-growth Douglas- fir forest. Agricultural and Forest Meteor- ology 63: 219–237. ———, S. C. SAUNDERS, T. R. CROW, R. J. NAIMAN, K. D. BROSOFSKE, G. D. MROZ, B. L. BROOKSHIRE, and J. F. FRANKLIN. 1999. Microclimate in forest ecosystem and landscape ecology. BioScience 49: 288–297. COBB, M. A., P. J. P. GOGAN, K. D. KOZIE, E. M. OLEXA, R. L. LAWRENCE, and W. T. ROUTE. 2004. Relative spatial distribution and habitat use patterns of sympatric moose and white-tailed deer in Voya- geurs National Park, Minnesota. Alces 40: 169–191. COLE, G. F. 1987. Changes in interacting spe- cies with disturbance. Environmental Management 11: 257–264. DANIELSON, E. W., J. LEVIN, and E. ABRAMS. 1997. Meterology. McGraw-Hill, New York, New York, USA. DEMARCHI, M. W., and F. L. BUNNELL. 1993. Estimating forest canopy effects on sum- mer thermal cover for Cervidae (deer family). Canadian Journal of Forestry Research 23: 2419–2426. ———, and ———. 1995. Forest cover selection and activity of cow moose in summer. Acta Theriologica 40: 23–36. DOU, H., G. JIANG, P. STOTT, and R. PIAO. 2013. Climate change impacts popula- tion dynamics and distribution shift of moose (Alces alces) in Heilongjiang pro- vince of China. Ecological Research 28: 625–632. DUSSAULT, C., J. P. OUELLET, R. COURTOIS, J. HUOT, L. BRETON, and J. LAROCHELLE. 2004. Behavioral responses of moose to thermal conditions in the boreal forest. Ecoscience 11: 321–328. DZIALOWSKI, E. M. 2005. Use of operative and standard operative temperature models in thermal biology. Journal of Thermal Biology 30: 317–334. ELLIS, C. R., and J. W. POMEROY. 2007. Esti- mating sub-canopy shortwave irradiance to melting snow on forested slopes. Hydrological Process 21: 2581–2593. ESRI. 2011. Version 9.3 User Manual. Envir- onmental Systems Research Institute, Redlands, California, USA. ERDAS Inc. 2010. ERDAS Field Guide. ERDAS, Inc., Atlanta, Georgia, USA. FABER-LANGENDOEN, D., N. AASENG, K. HOP, M. LEW-SMITH, and J. DRAKE. 2007. Vegetation classification, mapping, and monitoring at Voyageurs National Park, Minnesota: An application of the U.S. National Vegetation Classification. Applied Vegetation Science 10: 361– 374. GOGAN, P. J. P., K. D. KOZIE, E. M. OLEXA, and N. S. DUNCAN. 1997. Ecological status of moose and white-tailed deer in Voya- geurs National Park, Minnesota. Alces 33: 187–201. IPCC. 2007. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assess- ment Report of the Intergovernmental Panel on Climate Change. Geneva, Switzerland. JOHNSTON, C. A., and R. J. NAIMAN. 1990. Aquatic patch creation in relation to beaver population trends. Ecology 71: 1617–1621. 118 FOREST TEMPERATURE PATTERNS – OLSON ET AL. ALCES VOL. 50, 2014 KARNS, P. D. 2007. Population distrubution, density and trends. Pages 125-139 in A.W. Franzmann and C.C. Schwartz, edi- tors. Ecology and Management of the North American Moose. Second edition. Smithsonian Institution, Washington D.C., USA. KENNEDY, R., and A. KIRSCHBAUM. 2010. Standard operating procedure #4: Devel- oping LandTrendr maps of change. In Landsat-based monitoring of landscape dynamics in the National Parks of the Great Lakes Inventory and Monitor- ing Network (Version 1.0). National Resource Report NPS/GLKN/NRR- 2010/221. National Park Service, Fort Collins, Colorado, USA. ———, ———, U. GAFVERT, P. NELSON, Z. YANG, W. COHEN, E. PFAFF, and B. GHOLSON. 2010. Landsat-based monitoring of landscape dynamics in the National Parks of the Great Lakes Inventory and Monitoring Network (Version 1.0). National Resource Report NPS/GLKN/NRR-2010/221, National Park Service, Fort Collins, Colorado, USA. KIRSCHBAUM, A. A., and U. B. GAFVERT. 2010. Landsat-based monitoring of land- scape dynamics at Voyageurs National Park, 2002–2007. Natural Resources Technical Report NPS/GLKN/NRTR- 2010/356. National Park Service, Fort Collins, Colorado, USA. LECHOWICZ, M. J. 1984. Why do temperate deciduous trees leaf out at different times? Adaptation and ecology of forest communities. The American Naturalist 124: 821–842. LENARZ, M. S., M. E. NELSON, M. W. SCHRAGE, and A. J. EDWARDS. 2009. Tem- perature mediated moose survival in northeastern Minnesota. Journal of Wild- life Management 73: 503–510. LOOKINGBILL, T. R., and D. L. URBAN. 2003. Spatial estimation of air temperature dif- ferences for landscape-scale studies in montane environments. Agricultural and Forest Meteorology 114: 141–151. LOWE, S. J., B. R. PATTERSON, and J. A. SCHAE- FER. 2010. Lack of behavioral responses of moose (Alces alces) to high ambient temperatures near the southern periphery of their range. Canadian Journal of Zool- ogy 88: 1032–1041. MCCANN, N. P., R. A. MOEN, and T. R. HAR- RIS. 2013. Warm-season heat stress in moose (Alces alces). Canadian Journal of Zoology 91: 893–898. MCGRAW, A. M., R. A. MOEN, and L. G. OVER- LAND. 2012. Effective temperature differ- ences among cover types in northeast Minnesota. Alces 48: 45–52. MURRAY, D. L., E. W. COX, W. B. BALLARD, H. A. WHITLAW, M. S. LENARZ, T. W. CUSTER, T. BARNETT, and T. K. FULLER. 2006. Pathogens, nutritional deficiency, and climate influences on a declining moose population. Wildlife Monographs 166: 1–30. NOAA (NATIONAL OCEANIC AND ATMO- SPHERIC ADMINISTRATION). 2010. Cli- matological data for International Falls, Minnesota. National Climatic Data Cen- ter, Ashville, North Carolina, USA. OJAKANGAS, R. W., and C. L. MATSCH. 1982. Minnesota’s Geology. University of Minnesota Press, Minneapolis, Minne- sota, USA. PEEK, J. M. 2007. Habitat relationships. Pages 351-375 in A. W. Franzmann and C. C. Schwartz, editors. Ecology and Management of the North American Moose. Second edition. Smithsonian Institution, Washington D.C., USA. RABUS, B. 2003. The shuttle radar topography mission—a new class of digital elevation models acquired by spaceborne radar. Journal of Photogrammetry and Remote Sensing 57: 241–262. REIFSNYDER, W. E., G. M. FURNIVAL, and J. L. HOROWITZ. 1971. Spatial and temporal distribution of solar radiation beneath forest canopies. Agricultural Meteorol- ogy 9: 21–37. ALCES VOL. 50, 2014 OLSON ET AL. – FOREST TEMPERATURE PATTERNS 119 RENECKER, L. A., and R. J. HUDSON. 1986. Seasonal energy expenditures and ther- moregulatory responses of moose. Cana- dian Journal of Zoology. 64: 322–327. ———, and ———. 1990. Behavorial and thermoregulatory responses of moose to high ambient temperatures and insect harassment in aspen-dominated forests. Alces 26: 66–72. RODRIGUEZ, E., C. S. MORRIS, and J. E. BELZ. 2006. A global assessment of the SRTM performance. Photogrammetric Engi- neering and Remote Sensing 72: 249–260. ———, ———, ———, C. CHAPIN, J. M. MARTIN, W. DAFFER, and S. HENSLEY. 2005. An assessment of the SRTM topo- graphic products. Technical Report JPL D-31639, Jet Propulsion Laboratory, Pasadena, California, USA. SCHWARTZ, C. C., and L. A. RENECKER. 2007. Nutrition and energetics. Pages 441-478 in A.W. Franzmann and C. C. Schwartz, editors. Ecology and Manage- ment of the North American Moose. Sec- ond edition. Smithsonian Institution, Washington D.C., USA. TELFER, E. S. 1988. Habitat use by moose in southwestern Alberta. Alces 24: 14–21. VAN BEEST, F. M., and J. S. MILNER. 2013. Behavioural responses to thermal condi- tions affect seasonal mass change on a heat-sensitive northern ungulate. PloS one 8: 1–10. ———, B. VAN MOORTER, and J. M. MILNER. 2012. Temperature-mediated habitat use and selection by a heat-sensitive northern ungulate. Animal Behavior 84: 723–735. VAN WAGTENDONK, J. W., R. R. ROOT, and C. H. KEY. 2004. Comparisons of AVIRIS and Landsat ETM+ detection capabilities for burn severity. Remote Sensing of Envir- onment 92: 397–408. VERNON, H. M. 1930. The measurement of radiant heat in relation to human comfort. Journal of Physiology 70: 15. ———. 1932. The measurement of radiant heat in relation to human comfort. Jour- nal of Industrial Hygiene. 14: 95. ———. 1933. The estimation of solar radia- tion in relation to its warming effect on the human body. Quarterly Journal of the Royal Meterological Society 59: 239. Windels, S. K. 2014. 2014 Voyageurs National Park moose population survey report. Natural Resource Data Series NPS/VOYA/NRDS-2014/645. National Park Service, Fort Collins, Colorado, USA. 120 FOREST TEMPERATURE PATTERNS – OLSON ET AL. ALCES VOL. 50, 2014 FINE-CALE TEMPERATURE PATTERNS IN THE SOUTHERN BOREAL FOREST: IMPLICATIONS FOR THE COLD-DAPTED MOOSE STUDY AREA METHODS Results DISCUSSION ACKNOWLEDGEMENTS REFERENCES