1 FORAGE AND HABITAT LIMITATIONS FOR MOOSE IN THE ADIRONDACK PARK, NEW YORK Samuel Peterson1, David Kramer2, Jeremy Hurst2, Donald Spalinger3, and Jacqueline Frair1 1State University of New York College of Environmental Science and Forestry, 1 Forestry Drive, Syracuse, New York, USA 13210;2New York Department of Environmental Conservation, Division of Fish and Wildlife, 625 Broadway, Albany, New York, USA 12233; 3University of Alaska, 3211 Providence Drive, Anchorage, Alaska 99508, USA ABSTRACT: We used browse availability models to estimate the number of reproductive female moose (Alces alces) that could be supported during summer and winter in the predominantly forested 23,000 km2 Adirondack Park and Forest Preserve (Park) in northern New York State, USA. We developed allometric equations to predict available browse biomass for individual plants and subsequent biomass estimates in 6 major cover types to estimate the moose carrying capacity within the Park. Our model incorporated the differential availability and nutritional quality of woody browse species within each cover type and changes in local browsing intensity due to competing vegetation under two different foraging constraints – protein and digestible energy. We estimated the carrying capacity as 8 (protein constraint) and 135 × (energy constraint) greater in winter than summer. Spatially-explicit estimates of summer range capacity (Animal Use Days, AUD) based on the protein constraint correlated best with variation in local moose density derived from winter aerial surveys (R2 = 0.75, P < 0.01, n = 18). Protein availability was limiting in summer (AUD = 457 moose) with sparse patches of regenerating forest (< 20 years old) on privately-managed lands estimated to support 86% more moose than the dominant matrix of wetlands and mature mixed- deciduous forest. The small and patchy moose population in the Park reflects the relative scarcity of regenerating forest and optimal foraging habitat. Given statutory constraints of timber harvest in the majority of the Park, active forest management on private inholdings will play an outsized role in managing the moose population. ALCES VOL. 58: 1–30 (2022) Key words: adirondacks; Alces alces; AUD; crude protein; energy; forage; habitat; moose; range capacity. The northeastern United States has one of the largest regional populations of moose (Alces alces americana) in North America (Jensen et al. 2018). However, within that region are pockets of low density moose as in the Adirondack Park and Forest Preserve in New York State that essentially repre- sent the entire state population with density estimated as 0.03 moose/km2 (J. Hinton, SUNY College of Environmental Science and Forestry, unpublished data). In contrast, density in adjacent Maine, New Hampshire, and Vermont has been ≥ 0.3 moose/km2 in recent decades (Wattles and DeStefano 2011). Unlike in nearby states, moose in the Reserve have limited access to large-scale anthropogenic forest disturbance associated with timber harvesting (Hicks 1986). Moose require an abundance and dense concentration of quality woody browse to meet their nutritional requirements (Illius and Gorden 1987, Shipley et al. 1994), ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 2 and large-scale disturbances (fire histori- cally) that set back forest succession creates optimal foraging habitat for about 20 years (Peek 2007). In the northeastern United States, the natural, large-scale return interval of inland forests is 1000–7500 years, and more often involves localized winter dam- age (e.g., ice storms) in small patches rather than large swaths of blown-down trees (Lorimer and White 2003, Millward and Kraft 2004). Small scale canopy distur- bances from pathogens (e.g., hemlock woolly adelgid [Adelges tsugae]) and storms (i.e., winter blowdown) drive localized gap dynamics in the region (Runkle 1982, Seymour et al. 2002) producing diffuse patches of early seral vegetation. The local scale of these natural disturbances contrast with the more temporally and spatially pre- dictable disturbances arising from timber harvest operations in the contiguous com- mercial forests of Maine, New Hampshire, and Vermont. The geographic and population expan- sion of New England moose in the 1970– 1990s was associated with unprecedented clear-cutting in response to a regional spruce budworm (Choristoneura fumifer- ana) infestation that created a contiguous swath of regenerating forest/optimal forag- ing habitat (Bontaites and Gustafson 1993, Wattles and DeStefano 2011); moderate- high populations have been maintained through continual timber harvest (Dunfey- Ball 2019). Research points to moose use and preference of regenerating forest habitat and early successional browse year-round throughout this area (Thompson et al. 1995, Scarpitti et al. 2005, Bergeron et al. 2011, Millette et al. 2014). Understanding the relationship between availability and nutritional quality of forage resources is imperative to accurately assess the nutritional capacity of a landscape. Hobbs and Swift (1985) estimated the amount of food required to achieve a diet of specific quality (i.e., food supply), which, when divided by daily dry matter intake rates for the target animal, provides an estimate of Animal Use Days (AUD). They used this approach in Colorado to evaluate differences in habitat quality between burned and unburned forests for bighorn sheep (Ovis canadensis) and mule deer (Odocoileus hemionus) in comparison to traditional range supply models (Hobbs and Swift 1985). Hanley et al. (2012) expanded the AUD approach with their Forage Resource Evaluation System for Habitat (FRESH) with Sitka black-tailed deer (Odocoileus hemio- nus sitkensis) by using a linear-programming model to estimate the maximum amount of forage biomass that could be pooled from available forage types, while meeting speci- fied nutritional requirements under specific constraints including foraging time, bite size, and diet composition. Importantly, such estimates only pro- vide a “snapshot” carrying capacity of a given range. By assuming all available for- age was harvested from a given area at a given point in time, the model approximates the days a captive animal is maintained at a desired nutritional plane. Such instanta- neous estimates ignore plant-herbivore interactions, plant phenology, and dynamic metabolic requirements. As such, they are best interpreted as an index of habitat qual- ity and for comparing the quality of differ- ent forage types, cover types, treatments, or areas at a given point in time (Cook et al. 2016). Though current models are best used as indices, incorporating additional con- straints such as resource competition among forage species and accounting for uncer- tainty may yield estimates more realistic of field conditions (i.e., a multiplier effect; White 1983). One often ignored aspect of estimating range quality is the compounding of ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 3 uncertainty as estimates of available bio- mass are scaled up from individual plants to local plot or transect measures, and again to the level of specific cover types or study areas. Dismissal of this attribute of sampling range quality can lead to biased or overly confident estimates of differ- ences in range capacity across heteroge- neous landscapes. Precision may be grossly overestimated when considered at a single foraging level only, as when accounting for inter-plot variation in biomass while ignoring the precision associated with allometric predictions of the biomass available on a given plant (McWilliam et al. 1993). Monte Carlo (MC) simula- tions can provide a useful approach to account for error propagation across multi- ple scales or processes (Harmon et al. 2007), and have been successfully used to estimate CO2 uptake in pine forests (Bowler et al. 2012), carbon pools in sub- tropical forests (Conti et al. 2014), and nitrogen density in northeastern hardwood forests (Yanai et al. 2010). We applied Monte Carlo simulations to 1) scale-up estimates of available browse biomass for moose from individual plants to major cover types, and 2) estimated range capacity for moose that accounted for uncertainty in forage quality and forag- ing constraints during summer and winter. Our approach identified the extent to which each component, whether measured empir- ically in this study or drawn from the liter- ature, contributed to potential bias in and variance around range capacity estimates. We focused on digestible energy and pro- tein as the two most limiting nutritional factors (Moen 1995). Ultimately, we com- pared the value of different plants and cover types to identify potentially limiting factors of habitat and management of moose within the Adirondack Park and Forest Preserve. STUDY AREA The study area was delineated as the 23,500 km2 Adirondack Park and Forest Preserve of New York State, hereafter referred to as “Park” (43°57’08.9”N 74°16’57.5”W), of which ~45% is publicly- managed forest preserves interspersed with private inholdings (~55% of the landscape). The majority of public land is protected by Article XIV of the New York State Constitution as “Forever Wild” which pre- cludes resource extraction or development of any kind. Approximately 25% of privately owned lands are designated for resource management including timber harvest through state-regulated conservation ease- ments. Forest canopies in the region are dominated by American beech (Fagus gran- difolia), red maple (Acer rubrum), sugar maple (A. saccharum), yellow birch (Betula alleghaniensis), and paper birch (B. papyrif- era). Common conifer species include white pine (Pinus strobus), eastern hemlock (Tsuga canadensis), and balsam fir (Abies bal- samea). Elevation in the Park ranges from <50 m on the shore of Lake Champlain to >1600 m in the High Peaks (Lake Placid) area. During data collection, monthly pre- cipitation averaged 84.6 mm in May- September 2016, 84.2 mm in December 2016 – March 2017, and 128.3 mm in May- September 2017 (n = 17 weather stations; NOAA National Centers for Environmental Information). METHODS Plant Sampling We sampled woody species along stratified transects (n = 104; Fig. 1) proportional to the coverage of upland mixed forest (≥ 497 m elevation, n = 38), lowland mixed forest (< 497 m, n = 34), conifer forest (n = 13), and wetlands (n = 19) within the Park using the generalized classification of the Terrestrial Habitat Map produced by The Nature ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 4 Fig. 1. Study area in northeastern New York State showing public lands (light gray; Wild Forest, Wilderness, Primitive Use and Canoe Areas), private lands (white; Rural, Low, Moderate, Industrial and Intensive Use, Hamlets and Resource Management Areas), and water bodies (dark gray). Also indicated are locations where browse biomass was sampled to build allometric equations (stars), transects where browse components were measured in the field and biomass was predicted using allometric equations (black circles), and locations where browse nutritional samples were collected (white circles). ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 5 Conservancy for the Northeast US and Atlantic Canada (Ferree and Anderson 2013; Appendix 1). We sub-stratified wetland tran- sects into open wetland (n = 15) and forested wetland (n = 4) classes to capture vegetation patterns based on presence of mature trees within a given wetland. We placed 13 addi- tional transects in the Chateaugay Woodlands area in the northern region of the Park to capture timber harvests of known age (6–8 years old; unpublished data, J. Santamour, LandVest Inc.). Transect start points were random locations (Create Random points in ArcGIS v.10.4) within each cover type; three 2 × 4-m plots were spaced 50 m apart along a transect. To ensure that we adequately sampled the compositional and productive variation within a landcover type, random locations were spaced a minimum of 500 m from neighboring points. The direction of each transect was randomized using a ran- dom number generator (1–360) and modified as needed to ensure all sample plots fell within the same cover type. If we were unable to sample the same cover type within a given plot, we used another random point. We collected summer samples in August-September 2016 and 2017 to assess peak biomass, and winter samples in December 2016 – January 2017. Individual plants were sampled to represent the full range of size observed along the transects (Peterson et al. 2020). We measured the basal diameter of the main stem (10 cm above the substrate) for tree growth, and the tallest height, longest width, and perpendic- ular width (to calculate volume) for shrub growth per individual plant (Peterson et al. 2020). We clipped all twigs that fell within a 0.5–3.0 m height stratum to 8-mm diameter, a cutoff expected to provide a liberal esti- mate of available biomass for moose by including both saplings and mature trees with biomass within the stratum (Seaton et al. 2002). Clippings collected in summer included leaves and twigs, and winter clip- pings included twigs only for deciduous spe- cies and twigs plus needles for balsam fir. We separated and dried leaves and woody mass to a constant mass in a forced air oven at 90 ˚C for 24 h and weighed dried biomass to the nearest gram. Allometric Equations We developed allometric equations for the 14 species that comprised ≥95% of the woody diet in winter and summer within the Adirondack region to more efficiently quan- tify available moose browse (McInnes et al. 1992, Visscher et al. 2006, Peterson et al. 2020). We developed a series of candidate models to evaluate environmental and topo- graphical impacts on dried biomass per spe- cies. The dried mass (g) of each individual clipping was ln-transformed to reduce the spread of error and to achieve a normally distributed error structure. We included the environmental site covariates of canopy cover, elevation, percent slope, and aspect. We calculated canopy cover as the average proportion of overstory covering a sample plot from 4 measurements at each of the car- dinal directions with a convex densiometer. Elevation, percent slope, and aspect were derived from a 90-m resolution Digital Elevation Model (US Geological Survey) using ArcGIS (ESRI, Redlands, California, USA). Aspect was represented as a binary covariate with northwest values (0–40 degrees and 221–359 degrees) assigned as 0 and southeast values (41–220 degrees) as 1; southeastern facing slopes represent opti- mal plant growth conditions and increased species richness (Olivero and Hix 1998). After initial inspection of the relation- ship between stem size and biomass, we col- lapsed species (and size classes within species) into 11 groups prior to fitting the final models rather than analyzing the models per individual species based on ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 6 taxonomic similarities (Peterson et al. 2020; Table 1). All groups were analyzed for the summer and winter seasons except balsam fir which moose consume in winter and avoid in summer (Peterson et al. 2020). We evaluated candidate biomass models using all possible subsets with stepwise regression (Whittingham et al. 2006) and used Akaike’s Information Criterion (AICc) for final model selection (Burnham and Anderson 1998). All model covariates were centered by the mean and standardized prior to model fitting using a z-transformation (Schielzeth 2010). We calculated Pearson’s correlation coefficient for our predictors and developed candidate models with all plausi- ble covariate pairs having r < 0.7 (Dormann et al. 2012), as well as interactions between size and environmental site covariates. Additionally, we included quadratic covari- ates for basal diameter and volume to account for parabolic relationships in plant growth. In the case of uncertainty (i.e., where ∆AICc < 2.0), we predicted biomass with the Table 1. Allometric models predicting the browsable biomass available to moose in two consecutive winter and summer, 2016–17 as a function of lnbasal diameter (BD; for tree species only), volume (V; for bush species only), percent canopy cover (C), elevation in m (E), percent slope (S), and aspect (A). N indicates the number of samples within each group. Interactions among covariates are indicated by “:”. Group Species included N Season Model Adj. R2 Balsam fir Abies balsamea 17 Winter +8.39 + 0.88BD – 0.01E + 1.04A 0.88 Maples Sma Acer rubrum, A. saccharum 15 Summer –2.09 + 2.05BD 0.62 Winter –1.43 + 1.89BD – 0.13S 0.71 Maples Lgb A. rubrum, A. saccharum 17 Summer +14.53 – 0.02E 0.29 Winter +14.92 – 0.02E 0.32 Maples 3 A. pennsylvanicum, A. spicatum 16 Summer –8.25 + 6.75BD – 0.85BD2 – 3.51A + 0.31(BD2:A) 0.88 Winter –0.01 + 1.17BD – 13.79A + 4.12(BD:A) 0.85 Birches Betula populifolia, B. papyrifera, 27 Summer +0.7 + 3.66BD – 0.52BD2 – 0.02C – 0.79S + 0.17(BD:S) 0.84 B. alleghaniensis, Ostrya virginiana Winter –26.44 + 15.49BD – 1.85 BD2 + 0.2C – 0.11(BD:C) + 0.01(BD2:C) 0.83 Beech Smc Fagus grandifolia 9 Summer +0.84 + 1.53BD – 0.36S 0.69 Winter +0.46 + 1.57BD – 0.40S 0.69 Beech Lgd F. grandifolia 2 Summer 5.04 NA Winter 4.57 NA Hobblebush Viburnum lantanoides 19 Summer +4.44 + 0.89V + 0.05V2 0.97 Winter +2.96 + 0.96V 0.94 Poplars Populus grandidentata, 21 Summer –8.07 + 6.22BD – 0.71BD2 0.83 P. tremuloides Winter –8.98 + 6.26BD – 0.71BD2 0.86 Cherries Prunus serotina, P. pensylvanica 24 Summer +11.49 + 2.19BD – 0.04BD2 – 2.02S + 1.36(BD:S) – 0.23(BD2:S) 0.87 Winter –16.07 + 11.42BD – 1.47BD2 – 0.13S 0.79 Wild raisin Viburnum nudum cassinoides 21 Summer +9.11 + 0.99V – 0.01E – 0.02S 0.95 Winter +8.97 + 1.00V – 0.01E – 0.03S 0.94 aBasal diameter < 55 mm; b Basal diameter ≥ 55 mm; cBasal diameter < 60 mm; dBasal diameter ≥ 60 m. ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 7 highest ranked model with lowest AICc score that satisfied model assumptions. Model assumptions were examined using the Shapiro-Wilk test for normality, Durbin- Watson test for autocorrelation, and Breusch- Pagan test for heteroskedasticity. Estimating Range Capacity We estimated available browse biomass for each cover type with Monte Carlo (MC) sampling (1000 iterations per strata) to account for variation in predictions at each scaling strata (individual plants, plots, tran- sects, and cover types) using R statistical software Version 4.0.1 (R Core Team 2020). We were able to propagate uncertainties at multiple spatial scales by obtaining confi- dence intervals from the 1000 iterations per each scale. First, we applied allometric equa- tions to predict the available browse biomass on each individual plant by species. We mul- tiplied that prediction by the proportion of the browse on each individual plant species that fell within the defined sampling plot to account for incidences where the entire sam- pled plant did not fall within the plot. We then resampled predicted values using MC methods to create a grand mean for each individual plant. We summed the predicted biomass across all individuals of a species within each plot. Drawing from these values, a second MC sampling generated new val- ues of plot-level biomass for each species, from which we derived a mean biomass per species per plot. Across each transect, we summed values of biomass per species/plot and conducted a third MC sampling to gen- erate new values of transect-level biomass for each species. We summed predicted bio- mass values along transects per cover type, with a final MC sampling generating new values of cover type-level biomass for each species. We converted the final biomass esti- mates to kg/ha per species for inclusion in range capacity estimates for each cover type. We used the FRESH modeling tech- nique (Hanley et al. 2012) to quantify the range capacity of each forage and cover type for moose. We specified the model based on the energy and crude protein requirements of an adult female pregnant in winter and lac- tating in summer. To determine energetic requirements, the mean body weight (BW) for an adult female moose in this region was set to 350 kg based on conversion of field- dressed (carcass) BW reported in New Hampshire, Vermont, and Maine (unpub- lished data, K. Rines, New Hampshire Fish and Game Department) with the equation: Live Weight = Carcass Weight × 1.46 (Crichton 1997) (1) We derived values for daily weight loss, dry matter intake, metabolizable energy requirements (ME), foraging time, and bite size from the literature (Table 2). Estimates of crude protein (CP) and digestible energy (DE) available in principle forage species were available from a related study of moose diets in the Park (Peterson et al. 2020). The FRESH model used an energy constraint based on ME requirements, but required DE values for the input of different forage species; the relationship was assumed as ME ≈ DE × 0.82 (D. Spalinger, author). The FRESH model included an input called MAX that allowed the user to define the maximum amount of biomass of a browsed species that could be included in an individual’s diet (Felton et al. 2020). We defined this from the field observed utiliza- tion rate of each species from previously conducted browse selection surveys (Peterson et al. 2020). We calculated a rela- tive utilization index for each I species with the following equation (Gallant et al. 2004, Raffel et al. 2009, Harrison 2011): Ui = # browsed twigs / # available twigs (2) ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 8 Ui was quantified at the transect level and averaged for each species within each cover type. We further adjusted the available browse biomass value predicted for each cover type to account for changes in local moose browsing intensity (Fc) due to inter- fering vegetation (i.e., beech and conifer cover; Peterson et al. 2020). To do so, we calculated Fc at the cover type level by quan- tifying the average total biomass of principal browse species, percent cover of beech, and percent cover of conifer within each cover type, and determined the impacts on browse intensity with and without the effects of beech and conifer (Peterson et al. 2020). Fc was applied as a weight (summer = 0.81– 0.99, winter = 0.94–0.98) to the biomass value for each forage species by cover type in the FRESH model. The estimates of Animal Use Days (AUD) supported by available browse were ultimately compared with and without the adjustment of Fc. Applying point estimates for each component in the FRESH model, initial estimates of AUD were achieved under 2 alternative constraints: 1) a ME constraint set at 55,400 kJ/day and 27,827 kJ/day for summer and winter, respectively, and 2) a CP constraint set at 9.12 and 6.86% for sum- mer and winter, respectively (Regelin et al. 1985, Reese and Robbins 1994). For each scenario, we calculated total range capacity across the Park by multiplying the AUD/ha of each cover type by the areal extent of each type on the landscape. Comprehensive map- ping of regenerating forest was lacking for the Park. Therefore, we estimated its cover- age from field surveys that measured the proportion of regenerating forest within pri- vate inholdings that were classified a priori as conifer forest, upland deciduous/mixed forest, and lowland deciduous/mixed forest (Ferree and Anderson 2013); the proportions were 20, 21, and 4%, respectively. We divided the mean AUD by 180 days (length of the summer or winter season), multiplied this value by 0.2 to apply a cropping rate ensuring sufficient regeneration without Table 2. Nutritional requirements and foraging constraints set for a 350 kg, pregnant or nursing, female moose in the FRESH Cervid model used to estimate Animal Use Days per hectare (AUD/ha) by cover type in the Adirondack Park, New York. Constraint Season Value Formula Sources Daily weight loss Both 0.4 kg/day Renecker and Hudson 1989 Dry matter intake Summer 11,570 g/day 143 × BW0.75 McArt et al. 2010, Renecker and Hudson 1989 Winter 3075 g/day 38 × BW0.75 Metabolizeable energy Summer 55,400 kJ/day 0.82 × (835 kJ × BW0.75) Schwartz et al. 1988, Renecker and Hudson 1989, Dungan et al. 2010 Winter 27,827 kJ/day 0.82 × (124 kcal × BW0.75) × 0.29 kcal/kJ Crude protein Summer 9.12% Schwartz et al. 1987, VanBallenberghe and Miquelle 1990 Winter 6.86% Foraging time Summer 534 min/day Risenhoover 1986, Dungan et al. 2010 Winter 347 min/day Bite size Summer 1.5 g dry mass Moen 1995, Moen et al. 1997 Winter 1.0 g dry mass ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 9 long-term damage to the range (Allen et al. 1987), and ultimately considered the season, constraint, and cover types producing the lowest mean AUD as most limiting to moose in the system. Spatially-explicit estimates of AUD were determined by multiplying the estimated AUD/ha of a given cover type by the local proportion of that cover type within a defined area. Spatially-explicit estimates of moose density were available within a sample of 3 × 10 km blocks surveyed across the Park in winter 2016 (J. Hinton, SUNY-ESF, unpub- lished data). Survey blocks were excluded from consideration where reliable, on-the- ground information on regenerating timber cuts was unavailable, yielding 18 blocks for this comparison. We calculated AUD using the scenario indicated as most limiting for moose in each season. We also calculated total browse biomass available in each season within these 18 blocks. To quantify the strength of relationship between local AUD (or browse biomass) and moose density, we fit a linear model for each scenario (winter versus summer, protein versus energy constraint) and compared models by the variance explained. Finally, we applied the FRESH model under the most limiting conditions to predict potential changes in AUD for moose under 3 scenarios of plausible future landscape changes in the region: 1) warmer and drier conditions leading to 10% conversion of wetland areas (open and wooded wetland types; 33,147 ha) to lowland mixed forest, 2) increased development leading to 10% con- version of coniferous and deciduous forests (176,193 ha) to non-habitat on private lands, and 3) increased timber harvest on private lands converting 2% of deciduous/mixed forest (30,276 ha) and 1% of conifer forest (2481 ha) to regenerating forest, a reason- able estimate based on consultation with local forest managers. We ran 1000 MC sim- ulations in each scenario. RESULTS Allometric Models We sampled 11–32 (ave. = 21) individual plants per species in each season and fit allo- metric models to 11 distinct plant groupings (Appendix 2 and 3). With the exception of large beeches (n = 2, intercept only model) and large maples (n = 17, elevation as sole predictor), basal diameter or volume metrics (with linear or nonlinear terms) were import- ant predictors of available browse biomass of individual plants. We did not detect a dif- ference in allometric relationships for any species among sampling years (models including year effects ∆AICc > 2.0). All selected models met the assumptions of normally-distributed (SW = 0.90 – 0.98), independent (DW = 1.40 – 3.87), and homoscedastic errors (BP = 0.15 – 10.60; all P > 0.05), with the exception of the northern wild raisin (Viburnum nudum cassanoides) model in summer which indicated dependent errors (DW = 1.40, P = 0.04). In both sea- sons, all competing best models (∆AICc < 2.0) violated assumptions of either nor- mally distributed, independent, or homosce- dastic error. The first model that met all model assumptions was less supported (∆AICc = 2.20) than those with violations, but that model predicted values that differed from the best-supported model by ~1% only. Therefore, we chose to use the top model despite violating model assumptions. Available Browse Biomass Four species that were widespread in each cover type represented ~70% of available browse biomass in summer (Fig. 2); hobble- bush (Viburnum lantanoides, 27.7%) and striped maple (Acer pensylvanicum, 23.8%) provided ~50% of biomass with red maple (11.4%) and yellow birch (9.2%) combining for ~20%. Northern wild raisin (10.2%) and sugar maple (9.7%) also provided ~20% combined but were not distributed evenly ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 10 Fig. 2. Estimated density of browsable biomass (kg/ha) for moose during summer in the Adirondack Park, 2016–17. Estimates correspond to principal browse species only, those making up 95% of moose seasonal diets, with mean values reported by primary cover type. ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 11 across all cover types. In winter, balsam fir dominated browse availability (57.4%; Fig. 3) followed by striped maple (14.9%) and yellow birch (8.7%). Across cover types, the estimated reduction in browse intensity ranged from 2.3% (open wetland) to 18.5% (upland and lowland deciduous/mixed for- est) in summer, and 1.1% (conifer forest) to 5.5% (open wetland) in winter. By cover type, available browse biomass ranged from 87.0 (lowland deciduous/mixed forest) to 556.5 kg/ha (regenerating forest) during summer, and 146.7 (upland decidu- ous/mixed forest) to 376.0 kg/ha (wooded wetlands) during winter. Regenerating forest provided 2–6 × more available browse in summer than other cover types, and yielded 64% more browse than the dominant cover type (upland and lowland deciduous/mixed forest covering 63% of the landscape). In contrast, availability of winter browse bio- mass was more evenly distributed with wooded wetland (23.1%), open wetland (21.3%), and regenerating forest (18.5%) providing ~60% combined. Available browse within regenerating forest was dominated by yellow birch (38.4%) and several maple spe- cies (35.3%), whereas balsam fir (58.5– 93.7%) was dominant in wetland types. In both seasons, available browse biomass was more homogeneous in regenerating forest than in other cover types (CV = 0.42 in sum- mer and 0.23 in winter), as well as among transects in that type (CV = 0.13 in summer and 0.089 in winter). Predicted Range Capacity The distribution of cover types across the Park was dominated by deciduous/mixed forest (64% combined): 33.3% lowland deciduous/mixed forest (737,710 ha), 30.9% upland deciduous/mixed forest (685,396 ha), 10.4% conifer forest (230,043 ha), 10.2% forested wetland (225,280 ha), 4.6% regenerating forest (102,214 ha), and 4.4% open wetland (97,557 ha). The summer range supported fewer adult female moose (0.00–5.97 AUD/ha across cover types) than the winter range (0.23–26.54 AUD/ha) (Table 3). Only upland deciduous/mixed for- est and regenerating forest provided browse sufficient to meet the protein needs of lactat- ing moose in summer. Regenerating forest supported 6 × more AUD/ha than upland deciduous/mixed forest based on protein requirements. Moreover, browse within regenerating forest and wooded wetland pro- vided the most abundant sources of available energy in summer. Every habitat type pro- duced sufficient browse in winter to meet the protein and energy requirements of pregnant moose due to the high use of balsam fir and the lower daily energetic requirement in winter relative to summer. In winter, the highest AUD was predicted within wooded and open wetlands because of the abundance of balsam fir. We predicted the total available browse (under a protein constraint) in summer to support 457 ± 240 SD reproductive females when accounting for uncertainty (Table 4). Spatially-explicit estimates (within individ- ual aerial survey blocks, n = 18) of summer AUD/ha under the crude protein constraint explained a moderate and significant amount of variation in the observed moose density via aerial surveys (R2 = 0.75, P < 0.01; Fig. 4). Block-level density estimates for moose were less strongly related to AUD estimates under the summer energy con- straint (R2 = 0.45, P < 0.01), winter energy constraint (R2 = 0.36, P < 0.01), and winter protein constraint (R2 = 0.12, P = 0.15). Likewise, available browse biomass within each sampling block during summer showed a moderately strong relationship with esti- mated moose density in winter (R2 = 0.64, P < 0.01), whereas browse biomass in winter was a poor predictor of winter moose den- sity (R2 = –0.06, P = 0.90). ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 12 Fig. 3. Estimated density of browsable biomass (kg/ha) for moose during winter in the Adirondack Park, 2016–17. Estimates correspond to principal browse species only, those making up 95% of moose seasonal diets, with mean values reported by primary cover type. ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 13 Landscape Changes Scenarios We used summer protein constraint for our landscape change scenarios because it was identified as the most limiting factor for moose across season (summer and winter) and resource (crude protein and energy). In terms of model sensitivity, and in the absence of diet selectivity and ignoring model uncertainty, the potential reduction in browse intensity due to interfering vegeta- tion alone yielded a point estimate for sum- mer range capacity of 587 reproductive females. Accounting for diet selection, the estimates of summer AUD were reduced 82.2%, and accounting for interfering vege- tation further reduced the estimates by 11.4%; combined, these factors reduced AUD 84.3%. Ultimately, uncertainty in diet selectivity explained the greatest proportion of total variance in the final estimates (ave. CV = 22% across types and scenarios). Uncertainty in the estimated crude protein content of species or in predicted biomass availability by cover type contributed mini- mally to the variance in final estimates (<1% each). Removal of mature forest yielded the largest change in range capacity either by Table 3. The estimated number of seasonal Animal Use Days (AUD) per hectare (with standard deviation) for moose in 6 different cover types within the Adirondack Park, New York. Error from biomass availability, nutritional content of browse and diet selection habits of moose (Peterson et al. 2020) were incorporated through a series of Monte Carlo simulations. Cover type Summer AUD Winter AUD Protein Energy Protein Energy Conifer forest – 0.48 (0.52) 13.81 (11.14) 13.51 (11.17) Upland decid/mixed forest 0.28 (0.23) 0.70 (0.34) 5.74 (3.30) 3.16 (2.69) Lowland decid/mixed forest – 0.71 (0.34) 16.18 (14.47) 15.84 (14.03) Wooded wetland – 5.97 (5.05) 26.02 (21.43) 15.93 (15.10) Open wetland – 0.57 (0.46) 23.74 (21.04) 26.54 (22.90) Regenerating forest 1.95 (1.69) 5.61 (2.87) 11.28 (9.23) 0.23 (0.22) Table 4. Estimates of Park-wide nutritional carrying capacity of reproductive female moose in Adirondack Park, New York by (A) total abundance and (B) density. Values shown incorporate error from biomass availability, nutritional quality of forage and diet selection habits (Ui ) of moose. Browse intensity values (Fc) are also included in these estimates. A. Total moose Summer Winter Crude Protein Energy Crude Protein Energy Mean 456.5 3453.20 32897.70 26998.70 SD 239.6 1267.10 12341.80 11612.10 B. Moose per Km2 Mean 0.02 0.15 1.4 1.15 SD 0.01 0.05 0.53 0.49 Fig. 4. Predicted number of reproductive female moose supported by available summer browse (Animal Use Days) within 18, 3 × 10 km2 survey blocks (based on protein requirement) compared to the empirically-estimated density of moose in those blocks during winter 2016. ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 14 reducing capacity through conversion to non-habitat (development scenario) or increasing capacity through conversion to regenerating forest (timber harvest scenario; Fig. 5). In contrast, loss of wetlands (drier conditions scenario) or conversion of wet- lands to lowland deciduous/mixed forest (the only scenario that adds additional for- est) yielded minor impact (<5%) on range capacity. Models predicted that range capac- ity might increase by 30% in summer through harvest of 30,000 ha of mixed forest and 2400 ha of conifer forest. All landscape change scenarios predicted minimal reduc- tion in winter range capacity (< ~5%). DISCUSSION Our models indicate that the current and future moose population within the Park is constrained by summer protein availability associated with lack of early successional/ regenerating forest. The summer population estimate of ~455 reproductive females (under a protein constraint) was surprisingly similar to the total population estimate from winter aerial surveys (~700 moose; J. Hinton, SUNY College of Environmental Science and Forestry, unpublished data); more importantly, spatially-explicit esti- mates explained a substantial amount of variation in local moose density. We recog- nize that while leaves and stems of woody species dominate moose diets (Belovsky 1981), aquatic vegetation also provides mea- surable summer-fall forage when available (Crete and Jordan 1981). Because we did not account for aquatic vegetation in the models and wetlands are widespread and common in the Park (14.6% was classified as open or forested wetland), we presume that our esti- mates of summer range capacity are some- what low. Conversely, our unreasonably large estimate of winter range capacity (>25,000 animals or >1 animal/km2) was biased high because we did not account for selectivity (Pastor and Danell 2003) within species and age classes of browse (e.g., bal- sam fir foliage), effects of secondary Fig. 5. Predicted changes in the total number of reproductive female moose supported by available browse in the Adirondack Park under 3 landscape change scenarios. Scenarios represent drying conditions leading to a 10% conversion of wooded and open wetland habitat to lowland deciduous/ mixed forest, increased development leading to 10% conversion of conifer and deciduous/mixed forest to non-habitat, and increased timber harvest leading to 1% conversion of conifer forest and 2% conversion of deciduous/mixed forest to regenerating forest. Values indicate the percent change in the number of female moose relative to contemporary range conditions. ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 15 compounds on limiting consumption of bal- sam fir (Parikh et al. 2017, Nosko et al. 2020), and the influence of snow depth on browse availability (Visscher et al. 2006). However, that winter browse would need to be reduced 99% to match the limitation on summer range arguably reflects the simple interaction of an abundance of balsam fir with low nutrient requirements in the winter model. Although sample size per modeled vegetative class was smaller than desirable, we did assess variation in available biomass production and palatability across both spe- cies and size classes by separating our sam- ples by size and species when estimating AUD (Peterson et al. 2020). Despite certain limitations in our assumptions and sample sizes, we believe our data and analyses were sufficient to conclude that summer protein limitation is and will be the primary determi- nant of the stability and growth of the Park moose population. Where this moose population stands with respect to potential nutritional carrying capacity remains an open question beyond the lack of regenerating forest habitat. The persistent, low density population is consis- tent with its stable, yet limited availability of summer forage protein, and not necessarily reflective of low recruitment and survival. Ungulate populations near carrying capacity trend toward reduced productivity (Couturier et al 2009, Wam et al. 2010), and the 3-year calf:cow ratio (0.5 ± 0.9 SD) from limited aerial sampling in the Park (NYS DEC, unpublished data) is considered moderate productivity (Kuzyk et al. 2018). This popu- lation is currently protected from harvest and without a major predator, with brain- worm (Parelaphostrongylus tenuis) and liver fluke (Fascioloides magna) causing 12% of incidental natural mortality (K. Schuler, Cornell University College of Veterinary Medicine, unpublished data). Brainworm is somewhat of concern given the high deer density around most of the Park and the current low-moderate deer den- sity within. The primary objective was to determine if nutritional limitation for moose occurs in the Park and our models indicate that the moose population is under summer protein constraint due to lack of optimal foraging habitat associated with regenerating forest. In nearby moose populations inhabiting mostly commercial forestland (Maine, New Hampshire, Vermont), optimal foraging habitat is generated continuously and used year-round (Dunfey-Ball 2019). Numerous regional studies have identified use and preference of regenerating forest in winter, investigated and measured winter browsing damage on valuable, regenerating decidu- ous species, and documented use of mature coniferous forest when mobility and activ- ity is restricted temporarily in extreme win- ter conditions (Thompson et al. 1995, Scarpitti et al. 2005, Bergeron et al. 2011, Millette et al. 2014, Andreozzi et al. 2016). Thus, it is not surprising that constrained availability of summer biomass was cor- related positively with winter moose den- sity, and conversely, that winter browse biomass dominated by balsam fir was a poor predictor of winter moose density. Our inflated estimates of winter ADU reflect the high availability and presumed use of bal- sam fir in the model, and inadvertently minimized the effect of limited deciduous browse in regenerating forest that is pre- ferred winter forage. That we found restricted seasonal forage/nutrition and low moose population density were linked in the Park reflects the direct relationship between availability of year-round optimal foraging habitat and population abundance (Peek 2007). Importantly, conversion of mature forests to regenerating stands would increase optimal and preferred forage in both summer and winter. ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 16 Legal harvest (hunting) is the most com- mon tool for managing ungulate populations within their carrying capacity or to reduce over-populations. Instituting harvest of the small Park population is not feasible under current statutory constraints. Although the recent history of regional moose harvest is short-term (since the 1980s), harvest reduc- tions were implemented by the mid-2000s in New Hampshire and Vermont despite con- servative harvest rates in populations of mul- tiple thousands of animals. In these small states, the interrelationships of moderate-high moose density, winter tick parasitism, and climate change have reduced productivity in primary moose range despite commercial private forests continuously producing opti- mal foraging habitat (Dunfey-Ball 2019, Jones et al. 2019, Pekins 2020, DeBow et al. 2021). In contrast, The Adirondack Park has a low density moose population, lack of tim- ber harvest and optimal foraging habitat, and to date, unmeasured population impact from winter tick. Because the population is con- strained by lack of optimal foraging habitat, strategic forest management (cutting) to address this limitation is desirable, yet that is also prohibited by statute in much of the Park. The regional population explosion of moose in the 1970–1990s reflected unprece- dented harvest rates of spruce-fir forests in response to a spruce budworm infestation in the 1970–80s. Ironically, a trio of forest pests may provide a natural process to improve moose habitat in the Park on public lands through canopy openings in response to the current invasion by hemlock woolly adelgid, red pine scale (Matsucoccus resinosae), and again, the spruce budworm in the northeast- ern United States. ACKNOWLEDGEMENTS We would like to thank our partners at SUNY-ESF, NYSDEC, WCS and Cornell University for their assistance. This project would not have been possible without the cooperation of commercial foresters of the Adirondacks including Lyme Adirondacks, The Forestland Group and Molpus Woodlands. Additionally, we would like to thank K. Powers, D. Tinklepaugh, R. Rich, D. DeGroff, and R. Tam for their assistance with data collection and field support. Funding for this project was provided by SUNY-ESF, NYS-DEC (Federal Aid in Wildlife Restoration Grant W-173-G), and the American Wildlife Conservation Society. REFERENCES Allen, A. W., P. A. JordAn, and J. W. Terrell. 1987. Habitat Suitability Index Models: moose, Lake Superior Region. United States Fish and Wildlife Service Report 82(10.155). United States Department of the Interior, Fish and Wildlife Service Research and Development, Washington, DC, USA. Andreozzi, H. A., P. J. Pekins, and l. e. kAnTAr. 2016. Using aerial survey observations to identify winter habitat use of moose in northern Maine. Alces 52: 41–53. doi: 10.1007/BF00346984 Belovsky, G. E. 1981. Optimal activity times and habitat choice of moose. Oecologia 48: 22–30. Bergeron, d. H., P. J. Pekins, P. J., H. F. Jones, and W. B. leAk. 2011. Moose browsing and forest regeneration: a case study in northern New Hampshire. Alces 47: 39–51. BonTAiTes, k. M., and k. gusTAFson. 1993. The history and status of moose and moose management in New Hampshire. Alces 29: 163–167. BoWler, r. A. l., M. Freden, M. BroWn, and T. A. BlAck. 2012. Residual vegeta- tion importance to net CO2 uptake in pine-dominated stands following moun- tain pine beetle attack in British Columbia, Canada. Forest Ecology and Management 269: 82–91. doi: 10.1016/j. foreco.2011.12.011 ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 17 BurnHAM, k. P., and d. r. Anderson. 1998. Model Selection and Inference: A Practical Information-Theoretic Approach. Springer-Verlag, New York, New York, USA. conTi, g., n. Perez-HArguindeguy, F. QueTier, l. d. gorne, P. JAureguiBerry, g. A. BerTone, l. enrico, A. cucHieTTi, and s. diAz. 2014. Large changes in car- bon storage under different land-use regimes in subtropical seasonally dry forests of southern South America. Agriculture, Ecosystems and Environment 197: 68–76. doi: 10.1016/j.agee.2014. 07.025 cook, J. g., r. c. cook, r. W. dAvis, and l. l. irWin. 2016. Nutritional ecology of elk during summer and autumn in the Pacific Northwest. Wildlife Monographs 195: 1–81. doi: 10.1002/wmon.1020 couTurier, s., s. d. côTé, J. HuoT, and r. d. oTTo. 2009. Body-condition dynam- ics of a northern ungulate gaining fat in winter. Canadian Journal of Zoology 87: 367–378. doi: 10.1139/Z09-020 CreTe, M., and P. A. JordAn. 1981. Régime Alimentaire des Orignaux du Sud-Ouest Québécois pour les Mois d’Avril à Octobre. Canadian Field-Naturalist 95: 50–56. cricHTon, v. 1997. Hunting. Pages 617–653 in A. W. Franzmann and C. C Schwartz, editors. Ecology and Management of the North American Moose. Smithsonian Institution Press, Washington, DC, USA. deBoW, J., J. Blouin, e. rosenBlATT, c. AlexAnder, k. gieder, W. coTTrell, J. MurdocH, and T. donovAn. 2021. Effects of winter ticks and internal parasites on moose survival in Vermont, USA. Journal of Wildlife Management 85: 1423–1439. doi: 10.1002/jwmg.22101 dorMAnn, c. F., J. eliTH, s. BAcHerAcHer, c. BucHMAnn, g. cArl, g. cArre, J. r. gArciA MArQuez, B. gruBer, B. lAFourcAde, P. J. leiTAo, T. MunkeMuller, c. McCleAn, P. e. osBorne, B. reineking, B. scHroder, A. k. skidMore, d. zurell, and s. lAuTenBAcH. 2012. Collinearity: a review of methods to deal with it and a simulation study evaluating their perfor- mance. Ecography 36: 27–46. doi: 10.1111/j.1600-0587.2012.07348.x dunFey-BAll, K. R. D. 2019. Moose den- sity, habitat and winter tick epizootics in a changing climate. MS Thesis, University of New Hampshire, Durham, New Hampshire, USA. dungAn, J. D., L. A. sHiPley, and R. G. WrigHT. 2010. Activity patterns, forag- ing ecology, and summer range carrying capacity of moose (Alces alces shirasi) in Rocky Mountain National Park, Colorado. Alces 46: 71–87. FelTon, A. M., e. HolMsTroM, J. MAlMsTen, A. FelTon, J. P. g. M. croMsigT, l. edenius, g. ericsson, F. WideMo, and H. k. WAM. 2020. Varied diets, including broadleaved forage, are important for a large herbivore species inhabiting highly modified landscapes. Scientific Reports 10: 1904. doi: 10.1038/ s41598- 020-58673-5 Ferree, C., and M. G. Anderson. 2013. A Map of Terrestrial Habitats of the Northeastern United States: Methods and Approach. The Nature Conservancy, Boston, Massachusetts, USA. gAllAnT, d., c. H. BeruBe, T. TerMBlAy, and l. vAsseur. 2004. An extensive study of the foraging ecology of beavers (Castor Canadensis) in relation to habi- tat quality. Canadian Journal of Zoology 82: 922–933. doi: 10.1139/z04-067 HAnley, T. A., d. e. sPAlinger, k. J. Mock, o. l. WeAver, and g. M. HArris. 2012. Forage resource evaluation system for habitat-deer: an interactive deer habi- tat model. Technical Report PNW- GTR-858. United States Department of Agriculture, Forest Service, Pacific Northwest Research Station General, Portland, Oregon, USA. HArMon, M. e., d. l. PHilliPs, J. BATTles, A. rAssWeiler, r. o. HAll, and W. k. lAuenroTH. 2007. Quantifying ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 18 uncertainty in net primary production measurements. Pages 238–260 in T. J. Fahey and A. K. Knapp, editors. Principals and Standards for Measuring Primary Production. Oxford University Press, New York, New York, USA. HArrison, A. M. 2011. Landscape influences on site occupancy by beaver and resultant foraging impacts on forest composition and structure (Adirondack Mountains, NY, USA). MS Thesis, State University of New York College of Environmental Science and Forestry, Syracuse, New York, USA. Hicks, A. 1986. History and current status of moose in New York. Alces 22: 245–252. HoBBs, N. T., and D. M. sWiFT. 1985. Estimates of habitat carrying capacity incorporating explicit nutritional con- straints. Journal of Wildlife Management 49: 814–822. doi: 10.2307/3801716 illius, A. W., and I. J. gordon. 1987. The allometry of food intake in grazing ruminants. Journal of Animal Ecology 56: 989–99. doi: 10.2307/4961 Jensen, W. F., J. r. sMiTH, M. cArsTensen, c. e. Penner, B. M. Hosek, and J. J. MAskey. 2018. Expanding GIS analysis to monitor and assess North American moose distribution and density. Alces 54: 45–54. Jones, H., P. Pekins, l. kAnTAr, i. sidor, d. ellingWood, A. licHTenWAlner, and M. o’neAl. 2019. Mortality assessment of moose (Alces alces) calves during suc- cessive years of winter tick (Dermacentor albipictus) epizootics in New Hampshire and Maine (USA). Canadian Journal of Zoology 97: 22–30. doi: 10.1139/ cjz-2018-0140 kuzyk, g., i. HATTer, s. MArsHAll, c. ProcTer, B. cAdsAnd, d. lireTTe, H. scHindler, M. Bridger, P. sTenT, A. WAlker, and M. klAczek. 2018. Moose population dynamics during 20 years of declining harvest in British Columbia. Alces 54: 101–119. loriMer, C. G., and A. S. WHiTe. 2003. Scale and frequency of natural disturbances in the northeastern US: implications for early successional forest habitats and regional age distributions. Forest and Ecology Management 185: 41–64. doi: 10.1016/S0378-1127(03)00245-7 McArT, s. H., d. e. sPAlinger, W. B. collins, e. r. scHoen, T. sTevenson, and M. BucHo. 2009. Summer dietary nitrogen availability as a potential bot- tom-up constraint on moose in south-central Alaska. Ecology 90: 1400– 1411. doi: 10.1890/08-1435.1 McInnes, P. F., r. J. nAiMAn, J. PAsTor, and y. coHen. 1992. Effects of moose browsing on vegetation and litter of the boreal for- est, Isle Royale, Michigan, USA. Ecology 73: 2059–2075. doi: 10.2307/1941455 McWilliAM, A. l. c, J. M. roBerTs, o. M. r. cABrAl, M. v. B. r. leiTAo, A. c. l. decosTA, g. T. MAiTelli, and c. A. g. P. zAMPAroni. 1993. Leaf area index and above-ground biomass of terra firme rain forest and adjacent clearings in Amazonia. Functional Ecology 7: 310–317. doi: 10.2307/2390210 MilleTTe, T. l., e. MArcAno, and d. lAFloWer. 2014. Winter distribution of moose at landscape scale in northeastern Vermont: a GIS analysis. Alces 50: 17–26. MillWArd, A. A., and C. E. krAFT. 2004. Physical influence of landscape on a large-extent ecological disturbance: the northeastern North American ice storm of 1998. Landscape Ecology 19: 99–111. Moen, R. A. 1995. Moose energetics, forag- ing strategies and landscape effects: a spatially explicit simulation model. PhD dissertation, University of Minnesota, St. Paul, Minnesota, USA. Moen, R. A., J. PAsTor, and Y. coHen. 1997. A spatially explicit model of moose foraging and energetics. Ecology 78: 505–521. doi: 10.1890/0012-9658 (1997) 078[0505: ASEMOM]2.0.CO;2 nosko, P., k. roBerTs, r. knigHT, and A. MArcellus. 2020. Growth and chemical ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 19 response of balsam fir saplings released from intense browsing pressure in the boreal forests of western Newfoundland, Canada. Forest Ecology and Management 460: 117839. doi: 10.1016/j.foreco.2019. 117839 olivero, A. M., and D. M. Hix. 1998. Influence of aspect and stand age on ground flora of southeastern Ohio forest ecosystems. Plant Ecology 139: 177–187. doi: 10.1023/A:1009758501201 PArikH, g. l., J. s. ForBey, B. roBB, r. o. PeTerson, l. M. vuceTicH, and J. A. vuceTicH. 2017. The influence of plant defensive chemicals, diet composition, and winter severity on the nutritional condition of a free-ranging, generalist herbivore. Oikos 126: 196–203. doi: 10.1111/oik.03359 PAsTor, J., and K. DAnell. 2003. Moose- vegetation-soil interactions: a dynamic system. Alces 39: 177–192. 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, 2nd edition. University Press of Colorado, Boulder, Colorado, USA. Pekins, P. J. 2020. Metabolic and popula- tion effects of winter tick infestations on moose: unique evolutionary circum- stances? Frontiers in Ecology and Evolution 8:176. doi: 10.3389/fevo. 2020.00176 PeTerson, s., d. krAMer, J. HursT, and J. FrAir. 2020. Browse selection by moose in the Adirondack Park, New York. Alces 56: 107–126. r core TeAM. 2020. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. . rAFFel, T. r., n. sMiTH, c. corTrigHT, and A. J. gATz. 2009. Central place foraging by beavers (Castor canadensis) in a complex lake habitat. American Midland Naturalist 162: 62–73. doi: 10.1674/ 0003-0031-162.1.62 reese, E. O., and C. T. roBBins. 1994. Characteristics of moose lactation and neonatal growth. Canadian Journal of Zoology 72: 953–957. doi: 10.1139/ z94-130 regelin, W. L., C. C. scHWArTz, and A. W. FrAnzMAnn. 1985. Seasonal energy metabolism of adult moose. Journal of Wildlife Management 49: 388–393. doi: 10.2307/3801539 renecker, L. A., and R. J. Hudson. 1989. Ecological metabolism of moose in aspen-dominated boreal forest, central Alberta. Canadian Journal of Zoology 67: 1923–1928. doi: 10.1139/z89-275 risenHoover, K. L. 1986. Winter activity patterns of moose in interior Alaska. Journal of Wildlife Management 50: 727–734. doi: 10.2307/3800990 runkle, J. R. 1982. Patterns of disturbance in some old-growth mesic forests of eastern North America. Ecology 63: 1533–1546. doi: 10.2307/3800990 scArPiTTi, d., c. HABeck, A. r. MusAnTe, and P. J. Pekins. 2005. Integrating habi- tat use and population dynamics of moose in northern New Hampshire. Alces 41: 25–35. scHWArTz, C. C., M. E. HuBBerT, and A. W. FrAnzMAnn. 1988. Energy require- ments of adult moose for winter maintenance. The Journal of Wildlife Management 52: 26–33. doi: 10.2307/ 3801052 scHWArTz, C. C., W. L. regelin, and A. W. FrAnzMAnn. 1987. Protein digestion in moose. Journal of Wildlife Management 51: 352–357. doi: 10.2307/3801052 scHielzeTH, H. 2010. Simple means to improve the interpretability of regres- sion coefficients. Methods in Ecology and Evolution 1: 103–113. doi: 10.1111/j.2041-210X.2010.00012.x seATon, C. T. 2002. Winter foraging ecol- ogy of moose in the Tanana Flats and Alaska Range Foothills. MS Thesis, https://www.R-­project.org/ https://www.R-­project.org/ ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 20 University of Alaska Fairbanks, Fairbanks, Alaska, USA. seyMour, r. s., A. s. WHiTe, and P. g. deMAyAndier. 2002. Natural distur- bance regimes in northeastern North America- evaluating silvicultural sys- tems using natural scales and frequen- cies. Forest Ecology and Management 155: 357–367. doi: 10.1111/j.2041- 210X. 2010.00012.x sHiPley, l. A., J. e. gross, d. e. sPAlinger, n. T. HoBBs, and B. A. Wunder. 1994. The scaling of intake rate in mammalian herbivores. American Naturalist. 143: 1055–1082. doi: 10.1086/285648 THoMPson, M. e., J. r. gilBerT, g. J. MATulA Jr., and k. i. Morris. 1995. Seasonal habitat use by moose on managed forest lands in northern Maine. Alces 31: 233–245. visscHer, d. r., e. H. Merrill, d. ForTin, and J. l. FrAir. 2006. Estimating woody browse availability for ungulates at increasing snow depths. Forest Ecology and Management 222: 348–354. doi: 10.1016/j.foreco. 2005.10.035 WAM, H. K., O. HoFsTAd, and E. J. solBerg. 2010. Differential forage use makes car- rying capacity equivocal on ranges of Scandinavian moose (Alces alces). Canadian Journal of Zoology 88: 1179– 1191. doi: 10.1139/ Z10-084 WATTles, D. W., and S. desTeFAno. 2011. Status and management of moose in the northeastern United States. Alces 47: 53–68. WHiTe, R. G. 1983. Foraging patterns and their multiplier effects on productivity of northern ungulates. Oikos 40: 377–384. doi: 10.2307/3544310 WHiTTingHAM, M. J., P. A. sTePHens, r. B. BrAdBury, and r. P. FreckleTon. 2006. Why do we still use stepwise modelling in ecology and behaviour? Journal of Animal Ecology 75: 1182–1189. doi: 10.1111/j.1365-2656.2006.01141.x yAnAi, r. d., J. J. BATTles, A. d. ricHArdson, c. A. BlodgeTT, d. M. Wood, and e. B. rAsTeTTer. 2010. Estimating uncertainty in ecosystem budget calculations. Ecosystems 13: 239––248. doi: 10.1007/ s10021-010-9315-8 ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 21 APPENDICES Appendix 1: Reclassification of Ecosystems defined by The Nature Conservancy’s Terrestrial Habitat Map for the Northeastern US and Atlantic (Ferree and Anderson 2013). Ecosystems were assigned to either conifer, deciduous/mixed, wetland, wooded wetland or no-sampling based on ecosystem descriptions provided by Ferree and Anderson 2013. The deciduous/mixed class was further separated into upland and lowland forests, using a cutoff of 497 m in elevation. TNC class TNC ecosystem Reclassification Upland Acadian Low Elevation Spruce-Fir-Hardwood Forest Conifer Upland Acadian Sub-boreal Spruce Flat Conifer Upland Acadian-Appalachian Alpine Tundra No Sampling Upland Acadian-Appalachian Montane Spruce-Fir-Hardwood Forest Conifer Upland Agriculture No Sampling Upland Appalachian (Hemlock)-Northern Hardwood Forest: drier Deciduous/Mixed Upland Appalachian (Hemlock)-Northern Hardwood Forest: moist-cool Deciduous/Mixed Upland Appalachian (Hemlock)-Northern Hardwood Forest: typic Deciduous/Mixed Upland Central Appalachian Dry Oak-Pine Forest Deciduous/Mixed Upland Central Appalachian Pine-Oak Rocky Woodland Deciduous/Mixed Upland Developed No Sampling Upland Glacial Marine & Lake Mesic Clayplain Forest Deciduous/Mixed Upland Great Lakes Alvar No Sampling Upland Laurentian Acidic Rocky Outcrop No Sampling Upland Laurentian-Acadian Acidic Cliff and Talus No Sampling Upland Laurentian-Acadian Calcareous Cliff and Talus No Sampling Upland Laurentian-Acadian Calcareous Rocky Outcrop No Sampling Upland Laurentian-Acadian Northern Hardwood Forest: high conifer Deciduous/Mixed Upland Laurentian-Acadian Northern Hardwood Forest: moist-cool Deciduous/Mixed Upland Laurentian-Acadian Northern Hardwood Forest: typic Deciduous/Mixed Upland Laurentian-Acadian Northern Pine-(Oak) Forest Deciduous/Mixed Upland Laurentian-Acadian Pine-Hemlock-Hardwood Forest: moist-cool Deciduous/Mixed Upland Laurentian-Acadian Pine-Hemlock-Hardwood Forest: typic Deciduous/Mixed Upland Laurentian-Acadian Red Oak-Northern Hardwood Forest Deciduous/Mixed Upland North-Central Appalachian Acidic Cliff and Talus No Sampling Upland North-Central Appalachian Circumneutral Cliff and Talus No Sampling Upland Northeastern Interior Pine Barrens Conifer Upland Northern Appalachian-Acadian Rocky Heath Outcrop No Sampling Upland Open Water No Sampling Upland Shrubland/grassland; mostly ruderal shrublands, regenerating clearcuts No Sampling Wetland Boreal-Laurentian Bog: Isolated/small stream Wetland Wetland Boreal-Laurentian-Acadian Acidic Basin Fen: Undifferentiated Wetland Wetland Glacial Marine & Lake Wet Clayplain Forest: Undifferentiated Wetland Wetland Laurentian-Acadian Alkaline Conifer-Hardwood Swamp: Isolated Wooded Wetland Wetland Laurentian-Acadian Alkaline Conifer-Hardwood Swamp: Lake/pond: any size Wooded Wetland Appendix 1 (continued) ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 22 Appendix 1 (continued): Reclassification of Ecosystems defined by The Nature Conservancy’s Terrestrial Habitat Map for the Northeastern US and Atlantic (Ferree and Anderson 2013). Ecosystems were assigned to either conifer, deciduous/mixed, wetland, wooded wetland or no-sampling based on ecosystem descriptions provided by Ferree and Anderson 2013. The deciduous/mixed class was further separated into upland and lowland forests, using a cutoff of 497 m in elevation. TNC class TNC ecosystem Reclassification Wetland Laurentian-Acadian Alkaline Conifer-Hardwood Swamp: Smaller river floodplain/riparian Wooded Wetland Wetland Laurentian-Acadian Freshwater Marsh: Isolated Wetland Wetland Laurentian-Acadian Freshwater Marsh: Lake/pond: any size Wetland Wetland Laurentian-Acadian Freshwater Marsh: Smaller river floodplain/riparian Wetland Wetland Laurentian-Acadian Large River Floodplain: Acidic Swamp Wetland Wetland Laurentian-Acadian Large River Floodplain: Alkaline Conifer-Hardwood Swamp Wooded Wetland Wetland Laurentian-Acadian Large River Floodplain: Conifer-Hardwood Acidic Swamp Wooded Wetland Wetland Laurentian-Acadian Large River Floodplain: Floodplain Forest Wooded Wetland Wetland Laurentian-Acadian Large River Floodplain: Freshwater Marsh Wetland Wetland Laurentian-Acadian Large River Floodplain: Wet Meadow-Shrub Swamp Wetland Wetland Laurentian-Acadian Wet Meadow-Shrub Swamp: Isolated Wetland Wetland Laurentian-Acadian Wet Meadow-Shrub Swamp: Lake/pond: any size Wetland Wetland Laurentian-Acadian Wet Meadow-Shrub Swamp: Smaller river floodplain/ riparian Wetland Wetland North-Central Appalachian Acidic Swamp: Isolated Wetland Wetland North-Central Appalachian Acidic Swamp: Lake/pond: any size Wetland Wetland North-Central Appalachian Acidic Swamp: Smaller river floodplain/riparian Wetland Wetland North-Central Appalachian Large River Floodplain: Acidic Swamp Wetland Wetland North-Central Appalachian Large River Floodplain: Acidic Swamp Wetland Wetland North-Central Appalachian Large River Floodplain: Freshwater Marsh Wetland Wetland North-Central Appalachian Large River Floodplain: Rich Swamp Wetland Wetland North-Central Appalachian Large River Floodplain: Rich Swamp Wetland Wetland North-Central Appalachian Large River Floodplain: Wet Meadow-Shrub Swamp Wetland Wetland North-Central Interior and Appalachian Acidic Peatland: Undifferentiated Wetland Wetland North-Central Interior and Appalachian Rich Swamp: Isolated Wetland Wetland North-Central Interior and Appalachian Rich Swamp: Lake/pond: any size Wetland Wetland North-Central Interior and Appalachian Rich Swamp: Smaller river floodplain/riparian Wetland Wetland North-Central Interior Wet Flatwoods: Undifferentiated Wetland Wetland Northern Appalachian-Acadian Conifer-Hardwood Acidic Swamp: Isolated Wooded Wetland Wetland Northern Appalachian-Acadian Conifer-Hardwood Acidic Swamp: Lake/pond: any size Wooded Wetland Wetland Northern Appalachian-Acadian Conifer-Hardwood Acidic Swamp: Smaller river floodplain/riparian Wooded Wetland ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 23 Appendix 2: Model selection table for allometric equations describing browse biomass availability on individual tree and shrub species for moose in Adirondack Park, New York during summer. Models for large (>60 mm diameter) American beech are not displayed, as the small sample size warranted an intercept only model. Only candidate models carrying >1% of cumulative model weight are displayed. Small Maples Main effects Interactions df logLik AICc delta weight BD 3 –14.22 36.62 0.00 0.19 BD2 3 –14.40 36.99 0.38 0.16 BD+D 4 –12.82 37.65 1.03 0.11 BD2+D 4 –13.04 38.07 1.46 0.09 BD+S 4 –13.26 38.52 1.90 0.07 BD2+S 4 –13.36 38.73 2.11 0.07 BD+BD2 4 –14.09 40.18 3.56 0.03 BD+ES 4 –14.14 40.27 3.66 0.03 E+BD 4 –14.17 40.34 3.73 0.03 BD2+ES 4 –14.30 40.60 3.99 0.03 E+BD2 4 –14.35 40.70 4.08 0.02 E+BD+D 5 –12.08 40.83 4.21 0.02 E+BD2+D 5 –12.38 41.42 4.81 0.02 BD+BD2+D 5 –12.70 42.06 5.44 0.01 BD+D+S 5 –12.71 42.09 5.48 0.01 BD+D+ES 5 –12.81 42.29 5.67 0.01 BD2+D+S 5 –12.88 42.43 5.81 0.01 Large Maples E 3 –24.63 57.10 0.00 0.21 N 3 –25.38 58.61 1.51 0.10 CC+E 4 –23.76 58.85 1.75 0.09 E+BD 4 –24.01 59.36 2.26 0.07 E+N 4 –24.22 59.77 2.67 0.06 E+S 4 –24.59 60.51 3.41 0.04 2 –28.06 60.98 3.88 0.03 BD+N 4 –24.95 61.24 4.14 0.03 BD2+N 4 –24.95 61.24 4.14 0.03 CC+N 4 –25.02 61.36 4.26 0.03 BD2 3 –26.97 61.79 4.69 0.02 BD 3 –26.99 61.83 4.73 0.02 S+N 4 –25.32 61.98 4.88 0.02 S 3 –27.16 62.16 5.06 0.02 CC+E+N 5 –23.48 62.42 5.32 0.01 CC+E+BD2 5 –23.68 62.82 5.72 0.01 CC+E+BD 5 –23.69 62.83 5.73 0.01 CC+E+S 5 –23.71 62.88 5.78 0.01 E+BD2+N 5 –23.74 62.94 5.84 0.01 E+BD+BD2 5 –23.74 62.94 5.84 0.01 E+BD+N 5 –23.77 62.99 5.89 0.01 Appendix 2 (continued) ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 24 Appendix 2 (continued): Model selection table for allometric equations describing browse biomass availability on individual tree and shrub species for moose in Adirondack Park, New York during summer. Models for large (>60 mm diameter) American beech are not displayed, as the small sample size warranted an intercept only model. Only candidate models carrying >1% of cumulative model weight are displayed. Small Maples Main effects Interactions df logLik AICc delta weight CC 3 –27.67 63.19 6.09 0.01 Birches CC+BD+BD2+S BD × S 7 –30.16 80.21 0.00 0.09 CC+BD+BD2+S BD2 × S 7 –30.42 80.73 0.52 0.07 CC+BD+BD2 CC × BD 6 –32.86 81.93 1.72 0.04 CC+BD+BD2 CC × BD+CC × BD2 7 –31.43 82.75 2.54 0.03 CC+BD+BD2+D+S BD × S 8 –29.76 83.52 3.31 0.02 CC+E+BD+BD2 CC × E+CC × BD 8 –29.78 83.56 3.35 0.02 CC+BD+BD2 CC × BD2 6 –33.68 83.57 3.35 0.02 CC+E+BD+BD2+S CC × E+CC × S 9 –27.54 83.66 3.45 0.02 CC+BD+BD2+S+N BD × S 8 –29.88 83.76 3.55 0.02 CC+E+BD+BD2+S BD × S 8 –29.96 83.91 3.70 0.01 CC+BD+BD2+D+S BD2 × S 8 –29.96 83.92 3.70 0.01 CC+BD+BD2+S CC × S 7 –32.03 83.96 3.75 0.01 CC+E+BD+BD2+D+S CC × E+CC × S 10 –25.14 84.03 3.82 0.01 CC+BD+BD2+S BD × S+BD2 × S 8 –30.11 84.21 4.00 0.01 CC+BD+S BD × S 6 –34.01 84.23 4.01 0.01 CC+BD+BD2+S CC × S+BD × S 8 –30.12 84.24 4.02 0.01 CC+BD+BD2+S CC × BD2+BD × S 8 –30.12 84.25 4.04 0.01 CC+BD+BD2+S+N BD2 ×S 8 –30.14 84.29 4.07 0.01 CC+BD+BD2+S CC × BD+BD × S 8 –30.15 84.30 4.09 0.01 CC+E+BD+BD2+S CC × E+BD × S 9 –27.94 84.47 4.26 0.01 BD+BD2+S 5 –35.86 84.58 4.37 0.01 Hobblebush E+V+V2 5 1 11.8 0 0.12 E+V+V2 ExV 6 0.778801 12.3 0.5 0.09 E+V 4 0.740818 12.4 0.6 0.09 V 3 0.704688 12.5 0.7 0.08 V+V2 4 0.57695 12.9 1.1 0.07 CC+E+V+V2 6 0.367879 13.8 2 0.04 CC+V+V2 5 0.286505 14.3 2.5 0.03 E+V ExV 5 0.201897 15 3.2 0.02 CC+V 4 0.201897 15 3.2 0.02 A+A2+H 5 0.19205 15.1 3.3 0.02 A+A2+H 5 0.19205 15.1 3.3 0.02 A+A2+V 5 0.19205 15.1 3.3 0.02 E+V+V2 5 0.182684 15.2 3.4 0.02 H+V 4 0.157237 15.5 3.7 0.02 A+V 4 0.157237 15.5 3.7 0.02 A+H 4 0.157237 15.5 3.7 0.02 A+H 4 0.157237 15.5 3.7 0.02 Appendix 2 (continued) ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 25 Appendix 2 (continued): Model selection table for allometric equations describing browse biomass availability on individual tree and shrub species for moose in Adirondack Park, New York during summer. Models for large (>60 mm diameter) American beech are not displayed, as the small sample size warranted an intercept only model. Only candidate models carrying >1% of cumulative model weight are displayed. Small Maples Main effects Interactions df logLik AICc delta weight V+V2 4 0.157237 15.5 3.7 0.02 H+V+V2 5 0.157237 15.5 3.7 0.02 A+V+V2 5 0.157237 15.5 3.7 0.02 CC+E+V 5 0.149569 15.6 3.8 0.02 E+V+V2+D 6 0.142274 15.7 3.9 0.02 E+A+A2+V 6 0.142274 15.7 3.9 0.02 E+A+A2+H 6 0.142274 15.7 3.9 0.02 E+A+A2+H 6 0.142274 15.7 3.9 0.02 CC+E+V+V2 ExV 7 0.122456 16 4.2 0.01 E+A+V 5 0.116484 16.1 4.3 0.01 Aspens BD+BD2 4 –13.39 37.28 0.00 0.33 CC+BD+BD2 5 –12.53 39.06 1.77 0.14 BD+BD2+S 5 –12.98 39.95 2.67 0.09 E+BD+BD2 5 –13.35 40.70 3.42 0.06 BD+BD2+D 5 –13.39 40.77 3.49 0.06 CC+E+BD+BD2 6 –11.94 41.88 4.60 0.03 CC+BD+BD2+S 6 –12.19 42.39 5.10 0.03 CC+BD+BD2 CC × BD2 6 –12.36 42.72 5.44 0.02 CC+BD+BD2 CC × BD 6 –12.41 42.82 5.54 0.02 CC+BD+BD2+D 6 –12.53 43.05 5.77 0.02 BD+BD2+D+S 6 –12.76 43.52 6.23 0.01 BD+BD2+S BD2 × S 6 –12.97 43.94 6.66 0.01 E+BD+BD2+S 6 –12.97 43.94 6.66 0.01 BD+BD2+S BD × S 6 –12.97 43.95 6.67 0.01 Cherries BD+BD2+S+Ea BD × S+BD2 × S 8 –24.88 75.36 0.00 0.16 E+BD+BD2+S BD ×S+BD2 × S 8 –25.41 76.42 1.06 0.09 CC+E+BD+BD2+S BD × S+BD2 × S 9 –23.70 78.26 2.90 0.04 E+BD+BD2+S 6 –30.91 78.77 3.41 0.03 E+BD+BD2+S+N BD × S+BD2 × S 9 –23.99 78.83 3.47 0.03 BD+BD2+S 5 –32.84 79.01 3.65 0.03 E+BD+BD2+S+Ea BD × S+BD2 × S 9 –24.10 79.05 3.69 0.03 CC+E+BD+BD2+S CC × BD2+BD × S+BD2 × S 10 –21.34 79.60 4.24 0.02 BD+BD2+S BD × S+BD2 × S 7 –29.41 79.81 4.45 0.02 BD+BD2+D+S+Ea BD × S+BD2 × S 9 –24.64 80.13 4.77 0.01 BD+BD2+S+Ea BD × S+BD × Ea+BD2 × S 9 –24.72 80.30 4.94 0.01 BD+BD2+S+Ea+N BD × S+BD2 × S 9 –24.73 80.32 4.96 0.01 BD+BD2+S+Ea 6 –31.69 80.33 4.97 0.01 BD+BD2+S+Ea BD × S+BD2 × S+BD2 × Ea 9 –24.75 80.36 5.00 0.01 Appendix 2 (continued) ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 26 Appendix 2 (continued): Model selection table for allometric equations describing browse biomass availability on individual tree and shrub species for moose in Adirondack Park, New York during summer. Models for large (>60 mm diameter) American beech are not displayed, as the small sample size warranted an intercept only model. Only candidate models carrying >1% of cumulative model weight are displayed. Small Maples Main effects Interactions df logLik AICc delta weight CC+BD+BD2+S+Ea BD × S+BD2 × S 9 –24.81 80.48 5.12 0.01 E+BD+BD2+D+S BD × S+BD2 × S 9 –24.86 80.59 5.23 0.01 CC+E+BD+BD2+S CC × BD+CC × BD2 9 –24.87 80.59 5.23 0.01 CC+E+BD+BD2+S CC × BD+BD × S+BD2 × S 10 –21.98 80.88 5.52 0.01 Northen Wild Raisin E+V+S 5 –2.69 19.40 0.00 0.14 E+V+N 5 –3.07 20.10 0.75 0.09 CC+E+V 5 –3.27 20.50 1.15 0.08 CC+V 4 –5.54 21.60 2.20 0.05 CC+E+V+S 6 –1.98 21.90 2.56 0.04 CC+E+V+N 6 –2.00 22.00 2.61 0.04 E+V+S E × V 6 –2.00 22.00 2.62 0.04 E+V+N E × V 6 –2.11 22.20 2.84 0.03 CC+V CC × V 5 –4.18 22.40 2.98 0.03 E+V+S V × S 6 –2.22 22.40 3.06 0.03 E+V E × V 5 –4.26 22.50 3.13 0.03 E+V+S+N 6 –2.39 22.80 3.39 0.03 CC+E+V CC × V 6 –2.56 23.10 3.73 0.02 E+V+N E × V+V × N 7 –0.61 23.80 4.45 0.02 CC+V+S 5 –4.93 23.90 4.47 0.01 E+V+N V × N 6 –3.03 24.10 4.67 0.01 V+S+N V × N 6 –3.03 24.10 4.68 0.01 CC+E+V E × V 6 –3.08 24.20 4.78 0.01 CC+E+V CC × E 6 –3.27 24.50 5.15 0.01 CC+V+N V × N 6 –3.34 24.70 5.29 0.01 Small Beech BD+S 4 –8.435 34.9 0 0.14 S 3 –12.516 35.8 0.96 0.09 Intercept 2 –14.96 35.9 1.05 0.08 E 3 –12.771 36.3 1.47 0.07 BD+E 4 –9.527 37.1 2.18 0.05 BD 3 –13.411 37.6 2.75 0.04 BD+D 3 –14.667 40.1 5.26 0.01 CC 3 –14.827 40.5 5.58 0.01 ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 27 Appendix 3: Model selection table for allometric equations describing browse biomass availability on individual tree and shrub species for moose in Adirondack Park, New York during winter. Models for large (>60 mm diameter) American beech are not displayed, as the small sample size warranted an intercept only model. Only candidate models carrying >1% of cumulative model weight are displayed. Small Maples Main effects Interactions df logLik AICc Delta Weight BD+S 4 –15.11 42.21 0.00 0.39 BD 3 –18.36 44.89 2.68 0.10 BD+S BD × S 5 –14.37 45.40 3.19 0.08 BD+D 4 –16.76 45.51 3.30 0.07 E+BD+S 5 –14.79 46.25 4.04 0.05 BD+D+S 5 –15.03 46.73 4.52 0.04 BD+S+N 5 –15.10 46.88 4.66 0.04 BD+N 4 –17.53 47.06 4.84 0.03 BD+D+N 5 –15.22 47.11 4.90 0.03 BD+N BD × N 5 –15.58 47.83 5.62 0.02 E+BD 4 –18.14 48.27 6.06 0.02 BD+D BD × D 5 –16.30 49.27 7.06 0.01 S 3 –20.56 49.30 7.09 0.01 E+BD+D+N 6 –13.41 49.31 7.10 0.01 Large Maples E 3 –25.00 57.85 0.00 0.44 E+BD 4 –24.53 60.39 2.54 0.12 E+N 4 –24.94 61.21 3.36 0.08 E+S 4 –24.99 61.31 3.46 0.08 N 3 –26.86 61.56 3.71 0.07 2 –28.86 62.58 4.73 0.04 E+BD 5 –23.98 63.41 5.56 0.03 BD 3 –27.94 63.72 5.87 0.02 S 3 –28.00 63.85 6.00 0.02 BD+N 4 –26.47 64.28 6.43 0.02 E+BD+N 5 –24.52 64.49 6.64 0.02 E+BD+S 5 –24.52 64.49 6.64 0.02 S+N 4 –26.75 64.84 6.99 0.01 E+S+N 5 –24.93 65.32 7.47 0.01 Striped Maple BD+D BD × D 5 –18.85 53.70 0.00 0.47 BD+BD2+D BD × D 6 –17.60 56.53 2.83 0.11 BD+BD2+D BD2 × D 6 –17.60 56.54 2.84 0.11 BD+D+S BD × D 6 –18.64 58.60 4.91 0.04 BD2+D+S BD2 × D 6 –19.28 59.90 6.20 0.02 Birches CC+BD+BD2 CC × BD+CC × BD2 7 –34.00 87.90 0.00 0.17 CC+BD+BD2 CC × BD 6 –36.43 89.06 1.16 0.09 CC+BD+BD2+S BD × S 7 –35.04 89.98 2.08 0.06 CC+BD+BD2+D CC × BD 7 –35.63 91.15 3.25 0.03 CC+BD+BD2+S BD2 × S 7 –35.65 91.19 3.29 0.03 Appendix 3 (continued) ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 28 Appendix 3 (Continued): Model selection table for allometric equations describing browse biomass availability on individual tree and shrub species for moose in Adirondack Park, New York during winter. Models for large (>60 mm diameter) American beech are not displayed, as the small sample size warranted an intercept only model. Only candidate models carrying >1% of cumulative model weight are displayed. Small Maples Main effects Interactions df logLik AICc Delta Weight CC+E+BD+BD2 CC × BD+CC × BD2 8 –33.62 91.24 3.34 0.03 CC+BD+BD2 CC × BD2 6 –37.53 91.26 3.36 0.03 CC+BD+BD2+D CC × BD+CC × BD2 8 –33.70 91.40 3.50 0.03 CC+BD+BD2+N CC × BD+CC ×BD2 8 –33.70 91.41 3.51 0.03 BD+BD2+S BD × S 6 –37.82 91.85 3.95 0.02 CC+BD+BD2+S CC × BD+CC × BD2 8 –33.95 91.90 4.00 0.02 BD+BD2+S BD2 × S 6 –38.12 92.44 4.54 0.02 CC+BD+BD2+N CC × BD 7 –36.29 92.47 4.57 0.02 CC+BD+BD2+S CC × BD 7 –36.36 92.61 4.71 0.02 CC+E+BD+BD2 CC × BD 7 –36.43 92.74 4.84 0.01 CC+BD+BD2+D+S BD × S 8 –34.41 92.82 4.92 0.01 CC+BD+BD2+D CC × BD+BD2 × D 8 –34.42 92.84 4.93 0.01 CC+BD+BD2+S BD × S+BD2 × S 8 –34.46 92.92 5.02 0.01 CC+BD+BD2+S+N BD × S 8 –34.51 93.01 5.11 0.01 CC+BD+BD2+D CC × BD2 7 –36.70 93.30 5.40 0.01 Hobblebush V 3 –7.68 22.95 0.00 0.13 CC+E CC × V 6 –2.79 24.57 1.62 0.06 CC+E+V2 CC × V 7 –0.52 25.23 2.28 0.04 H 4 –7.20 25.25 2.30 0.04 E 4 –7.26 25.37 2.42 0.04 E+V2 5 –5.38 25.38 2.43 0.04 CC 4 –7.30 25.45 2.50 0.04 D 4 –7.44 25.74 2.79 0.03 CC CC × V 5 –5.60 25.82 2.87 0.03 CC+V2 CC × V 6 –3.65 26.31 3.35 0.03 H+V2 5 –5.85 26.31 3.36 0.03 V2+D 5 –5.99 26.59 3.64 0.02 E+H 5 –5.99 26.60 3.65 0.02 CC+E+V2 6 –4.22 27.44 4.49 0.01 CC+V2 CC × V2 6 –4.30 27.59 4.64 0.01 E+H+V2 6 –4.46 27.93 4.98 0.01 Aspens BD+BD2 4 –12.05 34.60 0.00 0.33 BD+BD2+D 5 –11.32 36.63 2.04 0.12 BD+BD2+S 5 –11.45 36.91 2.31 0.10 CC+BD+BD2 5 –11.82 37.65 3.05 0.07 E+BD+BD2 5 –12.04 38.07 3.48 0.06 BD+BD2+D BD2 × D 6 –10.83 39.67 5.07 0.03 BD+BD2+D BD × D 6 –10.89 39.79 5.19 0.02 Appendix 3 (continued) ALCES VOL. 58, 2022 ADIRONDACK MOOSE BIOMASS – PETERSON et al. 29 Appendix 3 (Continued): Model selection table for allometric equations describing browse biomass availability on individual tree and shrub species for moose in Adirondack Park, New York during winter. Models for large (>60 mm diameter) American beech are not displayed, as the small sample size warranted an intercept only model. Only candidate models carrying >1% of cumulative model weight are displayed. Small Maples Main effects Interactions df logLik AICc Delta Weight CC+BD+BD2+D 6 –11.16 40.32 5.72 0.02 E+BD+BD2+S 6 –11.21 40.41 5.82 0.02 BD+BD2+D+S 6 –11.23 40.46 5.86 0.02 CC+BD+BD2+S 6 –11.28 40.56 5.97 0.02 E+BD+BD2+D 6 –11.31 40.63 6.03 0.02 BD+BD2+S BD × S 6 –11.35 40.69 6.10 0.02 BD+BD2+S BD2 × S 6 –11.35 40.69 6.10 0.02 Cherries BD+BD2+S 5 –34.04 81.41 0.00 0.15 E+BD+BD2+S 6 –32.76 82.46 1.05 0.09 BD+BD2+S+E 6 –32.88 82.69 1.28 0.08 CC+E+BD+BD2+S CC × BD2 8 –28.80 83.21 1.79 0.06 BD+BD2+D+S 6 –33.67 84.28 2.87 0.04 CC+BD+BD2+S CC × BD2 7 –31.65 84.31 2.89 0.04 CC+E+BD+BD2+S 7 –31.67 84.35 2.93 0.03 BD+BD2+S BD2 × S 6 –33.82 84.57 3.16 0.03 BD+BD2+S+N 6 –33.87 84.69 3.27 0.03 CC+BD+BD2+S 6 –33.98 84.90 3.48 0.03 E+BD+BD2+S+N 7 –32.23 85.46 4.05 0.02 E+BD+BD2+S BD2 × S 7 –32.27 85.54 4.12 0.02 E+BD+BD2+S+E 7 –32.45 85.90 4.49 0.02 E+BD+BD2+D+S 7 –32.51 86.02 4.60 0.01 E+BD+BD2+S E × BD2 7 –32.53 86.05 4.64 0.01 CC+BD+BD2+S+E CC × BD2 8 –30.36 86.32 4.90 0.01 BD+BD2+S+E+N 7 –32.81 86.62 5.20 0.01 CC+BD+BD2+S+E 7 –32.83 86.66 5.24 0.01 BD+BD2+S+E BD2 × E 7 –32.84 86.67 5.26 0.01 BD+BD2+S+E BD2 × S 7 –32.86 86.72 5.31 0.01 BD+BD2+D+S+E 7 –32.87 86.74 5.33 0.01 Northern Wild Raisin E+V+S 5 –5.57 25.15 0.00 0.20 E+V+N 5 –6.58 27.15 2.01 0.07 CC+E+V 5 –6.85 27.71 2.56 0.06 E+V+S E × V 6 –5.02 28.03 2.89 0.05 CC+V 4 –8.83 28.16 3.01 0.04 CC+E+V+S 6 –5.14 28.28 3.13 0.04 E+V+S V × S 6 –5.18 28.37 3.22 0.04 E+V+S+N 6 –5.46 28.92 3.77 0.03 E+V E × V 5 –7.64 29.27 4.13 0.03 CC+V CC × V 5 –7.71 29.42 4.27 0.02 Appendix 3 (continued) ADIRONDACK MOOSE BIOMASS – PETERSON et al. ALCES VOL. 58, 2022 30 Appendix 3 (Continued): Model selection table for allometric equations describing browse biomass availability on individual tree and shrub species for moose in Adirondack Park, New York during winter. Models for large (>60 mm diameter) American beech are not displayed, as the small sample size warranted an intercept only model. Only candidate models carrying >1% of cumulative model weight are displayed. Small Maples Main effects Interactions df logLik AICc Delta Weight CC+E+V+N 6 –5.80 29.60 4.46 0.02 CC+V+S 5 –7.90 29.81 4.66 0.02 E+V+N E × V 6 –5.91 29.82 4.68 0.02 V+S+N V × N 6 –6.04 30.09 4.94 0.02 V 3 –11.41 30.22 5.08 0.02 V+S 4 –9.90 30.30 5.16 0.02 CC+E+V CC × V 6 –6.30 30.60 5.46 0.01 E+V+N E × V+V × N 7 –4.13 30.87 5.73 0.01 Small Beech BD+S 4 –8.914 35.8 0 0.14 S 3 –12.83 36.5 0.64 0.1 Intercept 2 –15.5 37 1.17 0.08 CC 3 –13.11 37 1.2 0.08 BD+CC 4 –10.04 38.1 2.25 0.05 BD 3 –14.1 39 3.17 0.03 D 3 –15.19 41.2 5.35 0.01