678 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 22001133 NORTNHorEthAeSaTsEteRrNn NNaAtTuUraRliAstLIST 2V0(o4l). :2607,8 N–6o9. 34 A Multi-scale Evaluation of the Effects of Forest Harvesting for Woody Biofuels on Mammalian Communities in a Northern Hardwood Forest David A. Patrick1,*, Cody Laxton1, Danielle Ball1, Scott Collins1, Stephanie Korzec1, Connor Langevin1, and Jonathan Vimislik1 Abstract - Research addressing the implications of forest harvesting for mammals has focused on different categories of silvicultural prescriptions. However, the effects of these prescriptions on forest structure can vary considerably, and categories of prescriptions rarely incorporate the market for which timber is being harvested. The latter information is important given the recent shift from conventional round-wood harvesting to wholetree removal for biofuels production, and corresponding reductions in post-harvest woody biomass left on-site. Our goal was to assess the effects of forest harvesting for biofuels on mammal species. Objectives included 1) evaluating how structural components influenced mammals, and 2) assessing the role of scale on species-habitat relationships. We sampled mammals in a 97-ha area of hardwood forest in the Adirondack Mountains in New York that had been partially harvested for biofuels in 2010. We used Sherman traps and track plates to assess the distribution of mammals. We captured 6 species of mammals in Sherman traps and identified 8 species using track plates. Mammalian species varied in their sensitivity to changes in habitat characteristics associated with biofuels harvest (coarse woody debris and slash). Our study reveals a complex suite of factors driving the response of mammals to variation in forest structure as a result of biofuels production. The harvesting practices used in the focal region are unlikely to lead to dramatic changes in the abundance and distribution of individual species of small mammals, but may influence the occurrence of common species including deer-mice and voles. Introduction The effects of modern silvicultural practices on forest structure span a wide range of intensities. At one end of the spectrum are techniques resulting in complete removal of mature trees (i.e., clearcutting) (Fuller et al. 2004). Conversely, techniques such as single-tree selection result in relatively little change to overall stand structure (Doyon et al. 2005). In general, forestry in the northeastern United States has seen a shift away from large-scale clearcutting towards a greater application of partial harvesting where some of the canopy is retained (Trani et al. 2001). From a wildlife perspective, one can argue that we need to move beyond evaluating the effects of silviculture based on the name of prescription alone (e.g., comparing clearcutting to partial harvesting), and instead focus on a more nuanced understanding of the consequences of harvesting on stand structure; the same harvesting technique may have very different implications depending on factors such as regional differences in the forest community and pre-harvest stand structure (Campbell et al. 2011, Duesser and Shuggart 1978). 1School of Natural Resources Management and Ecology, Paul Smith’s College, Paul Smiths, NY, 12970. *Corresponding author - dpatrick@paulsmiths.edu. Northeastern Naturalist Vol. 20, No. 4 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 679 Silviculture may influence mammalian habitat via changes in a number of characteristics: canopy cover, coarse woody debris (CWD); standing dead trees (snags); fine woody material (slash); the density of large trees, poles, and saplings; leaf-litter depth and composition; and the density of herbaceous vegetation in the understory (Bowman et al. 2000, Carey and Harrington 2001, Greenberg 2002, Loeb 1999). Mammal species exhibit a variety of responses to changes in these attributes depending on their ecology (Fisher and Wilkinson 2005, Sullivan et al. 2000, Zwolak 2009). For example, where forest harvesting creates conditions typical of early successional forest such as a more open canopy and a denser understory, mammal species adapted to these conditions are more likely to benefit (Litvaitis 2001). Conversely, if forest harvesting promotes attributes typical of mature forests, such as standing dead trees and dense canopy, a different mammalian community is likely to be promoted (Lomolino and Perault 2000, Sullivan et al. 2000). Silviculture does not necessarily equate to a given successional stage found in naturally regenerating forests. For example, forest harvesting techniques that maintain structural legacies found in mature forests such as snags and CWD may result in a stand structure that contains aspects of both early and later successional stages (Franklin et al. 2002). The effects of forest harvesting on mammalian communities are also scaledependent (Levin 1992, Orrock et al. 2000, Wheatley et al. 2005). For example, home-range locations may be related to fine-scale variation in habitat structure within a stand (Bowman et al. 2000, Loeb 1999), whereas population densities may be driven by the arrangement of patches of habitat at a larger landscape level (Van Horne 1983). These findings highlight the importance of quantifying the responses of mammals to variation in habitat structure at multiple scales when evaluating community assemblages (Durance et al. 2006, Riffelli et al. 2011, Saab 1999). Understanding how the different components and spatial distribution of forest structure influence mammalian communities is particularly relevant given changes in how harvested timber is used. To date, the majority of research focusing on silviculture and mammals has concentrated on the effects of conventional forestry prescriptions (e.g., Fuller et al. 2004, Zwolak 2009). However, the market for which timber is being harvested can directly influence harvesting practices on the ground, and hence, biodiversity. The consequences of changes in timber markets are particularly relevant given the current and projected shift from conventional harvesting (i.e., for saw-timber) towards harvesting forests for commercial biofuels (Galik et al. 2009, Nitterus et al. 2007). In biofuels harvesting, a much greater proportion of the tree is removed from the site through a process known as whole-tree chipping than occurs in conventional harvests (Hornbeck and Kropelin 1982, Stevens et al. 1995). This reduction in logging residue has been shown to alter forest structure and function, soil chemistry, and biotic community composition (Mahmood et al. 1999, Nitterus et al. 2007, Rudolphi and Gustafsson 2005). The goal of our study was to assess the effects of forest harvesting for biofuels on forest structure, and the distribution and abundance of mammal species in a northern hardwood forest. Specific objectives included 1) evaluating how structural components including canopy cover, CWD, slash, leaf litter, understory and 680 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 Northeastern Naturalist Vol. 20, No. 4 overstory structure influenced mammal presence and relative abundance; and 2) assessing the role of scale in influencing species-habitat relat ionships. Methods Study site Our study was conducted in a private forest located in the Adirondack Park in Northern New York State (44.4379˚N, 74.2695˚W). The 97-ha site encompassed two forest types existing in discontinuous stands: (1) Uneven-aged northern hardwood stands dominated by Fagus grandifolia Ehrh. (American Beech), Acer saccharum Marshall (Sugar Maple), and A. rubrum L. (Red Maple), primarily occupy east-facing slopes; and (2) even-aged hardwood-hemlock stands dominated by Tsuga canadensis L. (Eastern Hemlock), American Beech, Sugar Maple, Red Maple, and Betula alleghaniensis Britton (Yellow Birch) primarily occupy the west-facing slopes and wet areas. The current stand composition is a result of both natural and managed disturbances. The site was harvested in 1990 and managed as an even-aged system. In 1998, an ice storm caused minor crown damage in hardwood- hemlock stands. An improvement harvest focusing on northern hardwood stands was carried out in 2010 with the goal of removing all low-quality trees and salvaging trees infected with beech bark disease. The latter selective harvest was instrumental in creating the current unevenly aged hardwood stands. Both the 1990 and 2010 harvests involved whole-tree removal/chipping for biofuels production. Mammal and habitat sampling Nested sampling grids were established at two spatial scales in an attempt to reflect the relatively small and large home ranges of small- and medium-sized mammals (by medium we are referring to squirrels and some meso-carnivores). Initially, a large 6.25-ha grid (250-m x 250-m) was established along a baseline running southeast to northwest. This grid was subdivided into 36 nodes (6-m x 6-m) spaced at 50-m intervals, with these nodes used as the basis for placement of sampling arrays. We established 8 trapping grids to evaluate the relationship between patterns of small-mammal occurrence/relative abundance and habitat structure. We placed each grid at a randomly assigned node, and nodes had to be a minimum of 100 m apart. We centered our grids on the node; grids consisted of 16 Sherman live-traps (3 in x 3.5 in x 9 in) in a 4 x 4 pattern, with traps spaced at 10-m intervals. Each trap included a ball of synthetic nylon batting and a ball of peanut butter and oats as bait. Trapping occurred from 20–26 September 2012, with 4 of the 8 grids sampled on alternate nights such that each trap and grid was sampled for 3 nights. When checking a trap, we recorded whether or not it was sprung and if bait was present. If an animal was captured, we identified the species, age, and sex (Barnett and Dutton 1995) and noted whether or not the animal had been previously captured. On initial capture, we liberally marked each animal on the inside of the ear and nape of the neck with a permanent marker to avoid double-counting recaptures when estimating relative abundance (Barnett and Dutton 1995). Although these marks Northeastern Naturalist Vol. 20, No. 4 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 681 are temporary, prior research has indicated that retention time is sufficient for the duration of this study (D. Patrick, unpubl. data). We used track plates to assess the relationship between habitat structure and the occurrence of larger mammals. We assembled track plates using methods modified from Zielinski (1995), with each plate consisting of a 30.5-cm x 76-cm sheet of aluminum flashing sooted with a kerosene torch and covered with a 23-cm strip of adhesive contact paper at one end as a substrate for prints. Bait consisted of a mixture of peanut butter, oats, anchovies, and raw chicken. We placed this mixture in a folded section of hardware cloth, secured to the aluminum plate with heavy-gauge wire. We covered track plates with 3-mm clear-plastic sheeting held in place with wire to create an arched canopy approximately 30-cm high. The plastic also extended over the baited end of the plate, such that the only entrance to the track plate was from the front. We situated plates and canopy as close as possible to the node, with the back end abutting a large tree, debris pile, slope, or rock. A single trackplate was established at each of the 36 nodes on 1 October and collected 7 October 2012. To identify tracks, we removed the contact paper and evaluated the size and morphology of prints using a reference guide (Elbroch 2003). We also looked at the sooted portion of the track plate for additional prints. We used only those tracks that we could positively identify to species for further analyses. We quantified habitat structure independently at the trapping grid and node/ track-plate scale immediately following conclusion of our mammal sampling. For the trapping grid, we established a 2-m x 2-m plot centered on the location of each Sherman trap. Within each plot, we measured canopy cover using a spherical densiometer held at waist height directly above each node; the length and diameter of all downed CWD >10 cm in diameter; mean leaf-litter depth based on four samples taken in each of the corners of the plot; the percent cover of slash (<10 cm in diameter), vegetation <50 cm tall, and vegetation 51–100 cm tall; and the density of woody stems (<10 cm in diameter). At the node scale, we established a 5-m x 5-m plot surrounding each node. Within this plot, we measured the same variables as in our 2-m x 2-m plots, with two exceptions. To calculate downed CWD volume, we established four 10-m-long transects at each plot, one in each cardinal direction originating from the node. We then recorded the diameter at intersection and length of all logs >10 cm in diameter and within 2.5 cm of the ground intersecting each transect. These data were used to estimate CWD volume/ha using methods in Tierney and Faber-Langendoen (2009). We also measured the density of living trees (>10 cm dbh) within the larger plots. Statistical analyses All analyses were conducted using Program R version 2.15.2. (R Development Core Team, Vienna, Austria, 2006). We initially assessed correlation among independent variables for the three components of our analyses (individual trap, trap grid, and track plate). If variables were found to be significantly correlated (P ≤ 0.05), we removed one member of the pair to ensure only non-collinear variables were included in subsequent models. 682 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 Northeastern Naturalist Vol. 20, No. 4 To assess the factors driving presence/absence of small mammals at individual Sherman traps, we initially assessed if sampling effort influenced the likelihood of detecting a species (some traps malfunctioned leading to variation in the number of trap nights at each sampling location). To calculate the number of trap-nights at each sampling location, we excluded nights where the bait was removed but the trap not sprung, sprung traps without captures, and traps where animals had escaped from the rear of the trap. We then used logistic regression to assess if there was a relationship between trap nights and trap success. Because individual Sherman traps in the same grid were in close proximity, and therefore unlikely to be statistically independent, we assessed spatial dependence in captures of individual species among traps using packages “sp” and “spdep” in Program R (Plant 2012). We used presence or absence of an individual of the focal species at the trap as our response variable. Our spatial weights matrix was based on a maximum distance of 45 m, ensuring that traps within each of our 8 grids were considered neighbors. A two-tailed Moran’s I test was used for analysis of patterns of occurrence of each species because we could reasonably hypothesize both clustered patterns of occurrence and dispersed patterns due to competitive interactions. If no spatial dependence was found among captures, we developed a suite of candidate generalized linear models (GLMs) with a binomial distribution of errors to assess the relationship between presence/absence and habitat variables. Candidate models included a global model and models exploring suites of variables corresponding to different aspects of habitat structure. Variables in candidate models included the volume of CWD, leaf-litter depth (hereafter, litter), % canopy closure (hereafter, canopy), % slash cover (hereafter, slash), % vegetation 51–100 cm high (hereafter, veg_51_100), and the n woody stems <10-cm dbh/m2 (hereafter, stems). We then used Akaike’s information criterion (AIC) to select models that best explained variation in the data while minimizing the number of parameters (Burnham and Anderson 2002). If there were multiple models with a Δ AIC ≤ 2, we used multimodel inference to derive a weighted model-averaged estimate for each parameter (Mazerolle 2006). If spatial dependence among traps within the same grid was found, we employed autologistic models (Wintle and Bardos 2006). Under this approach, the program R function autocov_dist was initially used to create a new explanatory variable. This variable was then included in candidate models. Although autologistic regression models have received criticism, our application of a small neighborhood distance (45 m) is likely to have reduced this bias (Dormann et al. 2007, Plant 2012). To assess the factors driving the relative abundance of common focal species of small mammals across the focal area, we used trap success at a grid as our dependent variable and employed the same combination of candidate models and AIC as for the individual trap analyses. Trap success for each focal species was calculated as: n captures per grid/n trap nights. Trap-night calculations excluded traps if they were open with the bait removed, or if an animal had escaped from the back of the trap. We counted sprung traps without captures as a half trap-night, because some of these traps were likely to have sprung immediately following setting (and thus Northeastern Naturalist Vol. 20, No. 4 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 683 were unavailable for capturing animals), whereas some may have remained open throughout much of the night. Our independent variables for these analyses were calculated from the mean values for each of the habitat structure variables pooled across all the traps in the grid. To assess the role of habitat heterogeneity in driving presence/absence of larger species, we compared the presence of common species sampled at track plates (n = 36) to the habitat quantified at the track plates. Variables in candidate models included those listed for Sherman traps above with the addition of the number of large trees >10-cm dbh (hereafter, trees). Results We captured 6 small-mammal species in Sherman traps and positively identified 8 mammal species using track plates. The most common small-mammal species captured in Sherman traps were Peromyscus spp. (deer mice), Blarina brevicauda Say (Northern Short-tailed Shrew), and Myodes gapperi Vigors (Southern Redbacked Vole) (Table 1). We did not differentiate between the deer mice Peromyscus maniculatus Wagner (Deer Mouse) and P. leucopus Rafinesque (White-footed Mouse). Excluding small mammals, the most common species observed using track plates included Tamias striatus L. (Eastern Chipmunk), Tamiasciurus hudsonicus Erxleben (Red Squirrel), Sciurus carolinensis Gmelin (Eastern Gray Squirrel ), and Glaucomys sabrinus Shaw (Northern Flying Squirrel ) (Table 1). Our initial analysis of forest structure within the study area demonstrated that harvesting had resulted in a highly variable environment at multiple scales (Tables 2, 3, 4). The variability detected included patches ranging from early-successional conditions with an open canopy and dense understory, to patches reflecting characteristics of mature forest such as a closed canopy, mature trees, and little understory. We also observed considerable variation in specific factors associated with biofuels harvesting. For example, CWD varied from 0 to 164 m3/ha at the individual trap scale, and percent slash cover ranged from 1 to 90% (Table 2). Table 1. Mammals observed using Sherman traps and track plates in a partially-harvested northern hardwood forest in the northern Adirondack Park, NY, where timber was removed via whole-tree chipping for biofuels, September to October 2012. Species n captures/observationsA Sherman traps Peromyscus spp. 62 Myodes gapperi 14 Blarina brevicauda 14 Tamias striatus 8 Track platesA Tamias striatus 9 Tamiasciurus hudsonicus 5 Glaucomys sabrinus 3 ATracks of deer mice and voles were found on all track plates, but were excluded from analyses as the species could not be reliably distinguished from one another. 684 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 Northeastern Naturalist Vol. 20, No. 4 Three small-mammal species yielded sufficient captures for statistical analyses at individual Sherman traps: deer mice, Northern Short-tailed Shrew, and Southern Red-backed Vole. Following removal of traps with >1.5 trap nights from our dataset, we found no significant relationships between the probability of presence and sampling effort for these species (logistic regression; P > 0.35 for all species). We therefore included all traps with 2 or more trap nights in subsequent analyses. Initial evaluation also revealed a strong correlation between canopy openness and vegetative cover in the 0–50-cm class (linear regression; Table 4. Habitat sampled at 36 track plates during October 2012 in a partially-harvested northern hardwood forest in the northern Adirondack Park, NY, where timber was removed via whole-tree chipping for biofuels. Coarse woody-debris (CWD) was sampled using four 10-m long line-intersect transects. Vegetation and leaf litter was sampled within a 5-m x 5-m plot centered on each track plate. Variable Mean SE Range Volume of CWD (m3/ha) 66.705 10.295 4.442–323.776 Leaf-litter depth (cm) 6.0 0.5 0–15.0 % canopy closure 35 5 1–98 % slash cover 19 2 5–55 % vegetation 0–50 cm high 27 4 5–80 % vegetation 51–100 cm high 14 2 5–60 n woody stems/m2 (<10 cm dbh) 24 4 0–130 n trees/m2 (>10 cm dbh) 3 1 0–14 Table 3. Habitat sampled during October 2012 at eight trapping grids deployed in a partially harvested northern hardwood forest in the northern Adirondack Park, NY, where timber was removed via wholetree chipping for biofuels. Variable Mean SE Range Volume of CWD (m3/ha) 11.632 1.612 3.307–17.307 Leaf-litter depth (cm) 5.5 1.0 0–13.0 % canopy closure 40 11 3–79 % slash cover 18 4 5–75 % vegetation 0–50 cm high 34 10 0–35 % vegetation 51–100 cm high 12 4 5–40 n woody stems/m2 (less than 10 cm dbh) 40 15 6–130 Table 2. Habitat sampled during October 2012 at 128 plots in a partially-harvested northern hardwood forest in the northern Adirondack Park, NY, where timber was removed via whole-tree chipping for biofuels. Coarse woody-debris (CWD), slash, vegetation and leaf litter were sampled within a 2-m x 2-m plot centered on each trap. Variable Mean SE Range Volume of CWD (m3/ha) 17.516 2.589 0–164.402 Leaf-litter depth (cm) 5.7 0.3 0–15.0 % canopy closure 35 2 1–94 % slash cover 18 2 1–90 % vegetation 0–50 cm high 24 2 1–90 % vegetation 51–100 cm high 9 1 1–50 n woody stems/m2 (less than 10 cm dbh) 7 1 1–55 Northeastern Naturalist Vol. 20, No. 4 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 685 F1,127 = 67.180, P < 0.001), hence vegetative cover in this class was removed from subsequent models. Our assessment of spatial dependence demonstrated no significant relationship between presence of Northern Short-tailed Shrews and Southern Red-backed Voles and spatial location (Moran’s I = -0.051, P = 0.221; and Moran’s I = -0.004, P = 0.889, respectively). However presence of deer mice was influenced by the spatial location of the trap, with traps catching deer mice tending to be clustered together within a trapping grid (Moran’s I = 0.158, P = <0.001). The model that best explained the presence/absence of Northern Short-tailed Shrews included the density of woody stems (Table 5), with the presence of shrews decreasing as the density of stems increased (Table 6). For Southern Red-backed Voles, 4 models provided the best explanations for patterns in the data (Table 5), with the probability of vole presence decreasing as canopy cover increased, and Table 5. Generalized linear models relating the presence/absence of small-mammal species at individual Sherman traps to habitat attributes surrounding each trap. If spatial autocorrelation was found during preliminary analyses, an additional autocovariate term (ac) was included in subsequent models. n parameters Akaike weight Species/model (K) AICc ΔAIC (wi) Northern Short-tailed Shrews Stems 2 85.36 0.00 0.44 Litter 2 88.03 2.67 0.12 Canopy 2 88.11 2.75 0.11 Slash 2 88.29 2.93 0.10 CWD 2 88.50 3.14 0.09 Veg_51_100 2 88.43 3.06 0.09 CWD, slash, veg_51_100, stems 5 91.02 5.66 0.03 CWD, litter, slash 4 91.90 6.54 0.02 CWD, litter, canopy, slash, veg_51_100, stems 7 94.16 8.80 0.01 Southern Red-backed Voles Slash 2 78.23 0.00 0.22 Canopy 2 78.74 0.52 0.17 Stems 2 78.88 0.65 0.16 Veg_51_100 2 79.19 0.96 0.14 CWD 2 79.31 1.09 0.13 Litter 2 80.33 2.10 0.08 CWD, litter, slash 4 81.23 3.00 0.05 CWD, litter, canopy, slash, veg_51_100, stems 7 81.92 3.70 0.03 CWD, slash, veg_51_100, stems 5 82.19 3.96 0.03 Deer mice Stems, ac 3 144.68 0.00 0.19 Slash, ac 3 144.94 0.26 0.17 Litter, ac 3 145.09 0.41 0.16 Canopy, ac 3 145.12 0.44 0.15 CWD, ac 3 145.13 0.45 0.15 Veg_51_100, ac 3 145.13 0.45 0.15 CWD, litter, slash, ac 5 149.20 4.52 0.02 CWD, slash, veg_51_100, stems, ac 6 150.98 6.30 0.01 CWD, litter, canopy, slash, veg_51_100, stems, ac 8 155.48 10.80 0.00 686 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 Northeastern Naturalist Vol. 20, No. 4 increasing with greater amounts of slash, higher stem densities, and with greater vegetation cover (Table 6). Six models provided the best explanations for the presence/absence of deer mice (Table 6). However, all of the standard errors of parameter estimates for the habitat factors included in these models encompassed 0, indicating that these models had little predictive power. We had sufficient data to assess habitat factors driving relative abundance of deer mice and Northern Short-tailed Shrew at the trapping-grid scale. In order to meet assumptions of normality, deer mice abundances were square-root transformed (X’ = √X + 0.5) and Northern Short-tailed Shrew abundances were log transformed (X’ = log(X + 1)) (Zar, 1999). AICc scores indicated that our global models offered the best explanation for patterns in the relative abundance of these two species (Akaike weight [wi] = 1 for both species). However, all of the standard errors associated with parameter estimates for habitat factors in these models encompassed 0, indicating that these models had little predictive power. When analyzing presence of mammals at track plates, we focused on 4 species that were not sampled by Sherman traps: Northern Flying Squirrel, Eastern Gray Squirrel, Red Squirrel, and Eastern Chipmunk. Canopy cover, and the percent cover of vegetation in the 0–50-cm height class were significantly correlated (linear regression; F34 = 15.970, P = <0.001); hence, vegetation cover in this height class was removed from subsequent models. AIC scores indicated that the presence/absence of Northern Flying Squirrels was best explained by 6 competing models (Table 7). When interpreting parameter estimates derived from these models, however, only the number of large trees and the amount of CWD had standard errors that did not encompass zero (Table 8), with a higher probability of Northern Flying Squirrel presence in areas with more large trees and less CWD. Presence/absence of Gray Squirrels was best explained by CWD and Table 6. Parameter estimates for variables included in generalized linear models relating the presence/ absence of small-mammal species at individual Sherman traps to attributes of the habitat surrounding each trap. Species Parameter Model-averaged estimate Unconditional SE Northern Short-tailed Shrews Stems -0.339 0.246 Southern Red-backed voles Canopy -0.015 0.012 Slash 0.019 0.012 Stems 0.125 0.096 Veg_51_100 0.023 0.021 Deer mice Canopy 0.0009 0.0079 CWD -0.0002 0.0126 Litter -0.018 0.086 Slash -0.004 0.010 Stems 0.053 0.079 Veg_51_100 -0.001 0.017 Northeastern Naturalist Vol. 20, No. 4 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 687 Table 7. Generalized linear models relating the presence/absence of mammal species at 36 track plates to attributes of the habitat surrounding each sample. n parameters Akaike weight Species/model (K) AICc ΔAIC (wi) Northern Flying Squirrel Trees 2 22.76 0.00 0.20 CWD 2 22.79 0.03 0.19 Canopy, trees 3 22.96 0.20 0.18 Litter 2 24.05 1.29 0.10 Veg_51_100 2 24.28 1.52 0.09 Canopy 2 24.52 1.76 0.08 Stems 2 24.97 2.21 0.06 Slash 2 25.02 2.26 0.06 CWD, litter, slash 4 27.34 4.59 0.02 CWD, slash, veg_51_100, stems 5 29.18 6.42 0.01 CWD, litter, canopy, slash, veg_51_100, stems, trees 8 32.29 9.53 0.00 Gray Squirrel CWD 2 15.58 0.00 0.38 Veg_51_100 2 16.70 1.12 0.22 CWD, litter, slash 4 18.44 2.87 0.09 Slash 2 19.10 3.52 0.07 Stems 2 19.33 3.75 0.06 Litter 2 19.58 4.01 0.05 Trees 2 19.75 4.17 0.05 Canopy 2 19.76 4.18 0.05 CWD, slash, veg_51_100, stems 5 21.28 5.71 0.02 Canopy, trees 3 22.12 6.54 0.01 CWD, litter, canopy, slash, veg_51_100, stems, trees 8 28.49 12.92 0.00 Red Squirrel Canopy 2 27.24 0.00 0.26 CWD 2 27.98 0.75 0.18 Trees 2 29.16 1.92 0.10 Litter 2 29.18 1.94 0.10 Stems 2 29.36 2.12 0.09 Veg_51_100 2 29.26 2.02 0.09 Slash 2 29.48 2.24 0.08 Canopy, trees 3 29.63 2.39 0.08 CWD, litter, slash 4 32.79 5.55 0.02 CWD, slash, veg_51_100, stems 5 34.66 7.42 0.01 CWD, litter, canopy, slash, veg_51_100, stems, trees 8 41.09 13.85 0.00 Eastern Chipmunk Litter 2 37.77 0.00 0.34 Slash 2 38.29 0.52 0.26 CWD 2 39.61 1.84 0.13 CWD, litter, slash 4 40.77 3.00 0.08 Canopy 2 41.74 3.97 0.05 Veg_51_100 2 41.94 4.17 0.04 Stems 2 42.31 4.54 0.03 Trees 2 42.43 4.66 0.03 Canopy, trees 3 43.66 5.89 0.02 CWD, slash, veg_51_100, stems 5 43.72 5.94 0.02 CWD, litter, canopy, slash, veg_51_100, stems, trees 8 51.72 13.95 0.00 688 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 Northeastern Naturalist Vol. 20, No. 4 vegetation cover (Table 7), with a higher probability of presence in areas with more CWD and less vegetation (Table 8). Four models provided the best explanations for presence/absence of Red Squirrels (Table 7), however only canopy closure had a standard error not encompassing zero, with a higher probability of presence of Red Squirrels in areas with denser canopy (Table 8). Three models yielded competing explanations for the presence of Eastern Chipmunk (Table 7), with chipmunks more likely to be found in areas with more CWD, deeper leaf litter, and less slash (Table 8). Discussion The consequences of changes in forest structure due to biomass harvesting are of increasing concern to forest ecologists and managers (Riffelli et al. 2011). Unlike conventional round-wood harvesting, biomass harvesting via whole-tree chipping is likely to reduce the amount of fine woody material (slash) as well as CWD remaining on-site after harvest (Hornbeck and Kropelin 1982, Stevens et al. 1995). Although the effects of biomass harvesting on forest structure have been shown to vary depending on the exact silvicultural prescription (Riffelli et al. 2011), reductions in slash have been linked to a variety of effects on biodiversity including loss of forest-associated invertebrate species (Nitterus et al. 2007), increased physiological stress and decreased survival of amphibians (Rittenhouse et al. 2008), and changes in small-mammal community structure (Gunther et al. 1983, Hooven and Black 1976). Table 8. Parameter estimates for variables included in generalized linear models relating the presence/ absence of small-mammal species at trapping grids to attributes of the habitat surrounding each grid. Model-averaged estimates for parameters were calculated for mod els with Δ AICc < 2. Species Parameter Model-averaged estimate Unconditional SE Northern Flying Squirrel Canopy 0.030 0.030 CWD -0.027 0.023 Litter -0.024 0.024 Veg_51_100 -0.049 0.066 Trees 0.270 0.170 Gray Squirrel CWD 0.018 0.010 Cover_51_100 -0.239 0.209 Red Squirrel Canopy 0.028 0.190 CWD -0.016 0.016 Litter -0.119 0.214 Trees -0.121 0.247 Eastern Chipmunk CWD 0.011 0.007 Litter 0.459 0.245 Slash -0.112 0.070 Northeastern Naturalist Vol. 20, No. 4 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 689 In contrast with conventional forest harvesting, where post-harvest forest structure has been studied, less attention has been focused on quantifying the effects of biomass harvesting for forest structure following actual harvests, and most studies have relied on experimentally manipulated conditions (Riffelli et al. 2011). Our study demonstrated that the application of whole-tree chipping led to a highly heterogeneous landscape at multiple spatial scales. For example, we found that at a fine scale (comparing among samples taken at 10-m intervals), the amounts of both slash and CWD varied considerably, with some areas almost devoid of woody material, whereas other locations contained high densities. Directly comparing the amounts of woody material at our study site with prior research is difficult due to inherent differences in forest structure; however, the mean amount of CWD quantified at the individual-trap level at our site (17.5 ± 2.6 m3/ha) was approximately half that quantified in clearcuts in mature mixed forest in Maine where CWD was deliberately retained (45.6 ± 21.6 m3/ha), and slightly less than was found in control forests in the same study (22.9 ± 11.8 m3/ha) (Patrick et al. 2006). The mean percent coverage of slash in our sites (18 ± 1.74 %) was comparable to that found in partial harvests (16 ± 2.5 %) and clearcuts (20 ± 1.7 %) in Maine (D. Patrick, unpubl. data). These results suggest that biofuels harvesting may not result in dramatic reductions in the amount of remaining woody material on the forest floor compared to conventional harvests. We found similarities and differences among the responses of mammalian species to habitat change in our study area compared to prior research findings. In general, the composition of the mammalian community we described in our study is typical for hardwood forests in the focal region (Carey and Harrington 2001, Fuller et al. 2004, Glennon and Porter 2007). The lack of strong linkages between deer mice and forest structure at our site is supported by data indicating these are generalist species that occupy a wide variety of forest types (Bellows et al. 2001, Fisher and Wilkinson 2005, Fuller et al. 2004, Klenner and Sullivan 2009). Conversely, Southern Red-backed Voles have been shown to be more sensitive to habitat alteration, preferring moist habitat, and areas with more slash and less CWD (Orrock et al. 2000). We found a similar increase in the presence of voles in areas with more slash, but no relationship with CWD. We also found that Southern Red-backed Voles were generally associated with early-successional conditions including an open canopy and dense understory. Similar to deer mice, Northern Short-tailed Shrews are considered to be habitat generalists (Getz et al. 2004, Glennon and Porter 2007). Prior research in the Adirondacks supports this characterization, with occurrence of the species linked to both mature forest conditions including a closed canopy and more CWD, and characteristics of early-successional forest including increasing shrub abundance (Glennon and Porter 2007). We did not find a linkage between the presence of Northern Short-tailed Shrew and characteristics of mature forest such as CWD, but our findings did support the association of this species with vegetative cover in the 51–100-cm height category (i.e., shrubs associated with early-successional habitat). The sciurids we observed at track plates included species considered habitat generalists, such as the Eastern Red Squirrel, Gray Squirrel, and Eastern Chipmunk, 690 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 Northeastern Naturalist Vol. 20, No. 4 and species associated with mature forests, specifically Northern Flying Squirrel (Carey 1995, Glennon and Porter 2007). At our site, both Gray Squirrels, and Eastern Chipmunks tended to occur in areas with more CWD. These results differ from Glennon and Porter’s (2007) assertion that, in our region, these two species prefer areas with human development, but our results may well indicate the microhabitat features these animals select for when more disturbed landscapes are not present. The increase in the probability of Northern Flying Squirrel presence with increased density of large trees likely reflects a greater availability of tree cavities used by this species (Carey et al. 1992), and potentially a greater abundance of the hypogeous sporocarps of mycorrhizal fungi that form an important food source for this species (Waters and Zabel 1995). Our results also support the contention that it is important to consider multiple scales when understanding organism-habitat relationships (Levin 1992, Orrock et al. 2000). We showed that organisms were associated with different aspects of the habitat at different scales. For example, Southern Red-backed Voles were sensitive to fine-scale habitat variation, but not to variation in habitat structure at the grid scale. While these results may appear to be contradictory, they likely reflect the different drivers of patterns of habitat selection at the population level rather than the individual scale; individual voles selected microsites with more groundcover, but groups of voles did not demonstrate significant habitat selection at the larger grid scale. Our study was specifically designed to inform forest-management practices where whole-tree chipping leads to a reduction of post-harvest woody biomass on the forest floor, including both slash and CWD. The Southern Red-backed Vole was the only species positively linked to the amount of slash at the individual- trap scale. None of the small-mammal species we studied was associated with CWD, although two sciurid species showed a positive association with this habitat attribute. In general, our results do not suggest that the levels of variation in slash and CWD we observed in our study are likely to lead to dramatic changes in the mammalian community. Although our study offered valuable information relating the effects of wholetree chipping and variation in forest structure to mammalian community structure, there are a number of important caveats when interpreting our results. Most importantly, our analyses and conclusions are limited to a single study-site during the fall. Given the likelihood of regional and seasonal differences in patterns of habitat selection, care should be taken when extending our results to a larger spatial and temporal extent. Furthermore, our study design did not consider the larger landscape context that may influence patterns of distribution of more wide-ranging species including some of the sciurids we sampled. Bearing in mind these caveats, our study reveals a complex suite of factors driving the response of mammals to variation in forest structure as a result of partial harvesting for biofuels production. When making management recommendations for future forest harvesting based on the results of both our own study and prior research, we draw attention to the need to explicitly state management objectives. While harvesting practices such as those used in the focal region are unlikely to Northeastern Naturalist Vol. 20, No. 4 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 691 lead to dramatic changes in the abundance and distribution of individual smallmammal species, they are likely to influence the occurrence of common species including Southern Red-backed Vole. Acknowledgments We would like to thank Alisha Benack, Sam Forlenza, Richard Franke, Alec Judge, and Elena Zito for help with fieldwork, and Dan Bogan for advice on study design. We would like to extend particular thanks to Northwoods Forest Consultants, LLC and Herb Boyce for providing information and advice regarding the harvesting conducted on the site. Literature Cited Barnett, A., and J. Dutton. 1995. Expedition Field Techniques: Small mammals (excluding bats). Royal Geographical Society, London, UK. 131 pp. Bellows, A.S., J.F. Pagels, and J.C. Mitchell. 2001. Macrohabitat and microhabitat affinities of small mammals in a fragmented landscape on the upper coastal plain of Virginia. American Midland Naturalist 146:345–360. Bowman, J.C., D. Sleep, G.J. Forbest, and M. Edwards. 2000. The association of small mammals with coarse woody debris at log and stand scales. Forest Ecology and Management 129:119–124. Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Multi-Model Inference: A Practical Information-Theoretic Approach. 2nd Edition. Springer, New York, NY. 488 pp. Campbell, S.P., D.A. Patrick, and J.P. Gibbs. 2011. Biodiversity, conservation biology, and forest health. Pp. 277–320, In J.D. Castello and S.A. Teale, (Eds.). Forest Health. Cambridge University Press, Cambridge, UK. 392 pp. Carey, A.B. 1995. Sciurids in Pacific Northwest managed and old-growth forests. Ecological Applications 5:648–661. Carey, A.B., and C.A. Harrington. 2001. Small mammals in young forests: Implications for management for sustainability. Forest Ecology and Management 154:289–309. Carey, A.B., S.P. Horton, and B.L. Biswell. 1992. Northern Spotted Owls: Influence of prey base and landscape character. Ecological Monographs 62:223–250. Dormann, C.F., J.M. McPherson, M.B. Araujo, R. Bivand, J. Bollinger, C. Gudrun, R.G. Davies, A. Hirzel, W. Jetz, W.D. Kissling, I. Kuhn, R. Ohlemuller, P.R. Peres-Neto, B. Reineking, B. Schroder, F.M. Schurr, and R. Wilson. 2007. Methods to account for spatial autocorrelation in the analysis of species distributional data: A review. Ecography 30:609–628. Doyon, F., D. Gagnon, and J. Giroux. 2005. Effects of strip- and single-tree selection cutting on birds and their habiat in a southwestern Quebec northern hardwood forest. Forest Ecology and Management 209:101–115. Duesser, R.D., and H.H. Shuggart, Jr. 1978. Niche pattern in a forest floor small-mammal fauna. Ecology 60:108–118. Durance, I., C. Lepichon, and S.J. Ormerod. 2006. Recognizing the importance of scale in the ecology and management of riverine fish. River Research and Applications 22:1143–1152. Elbroch, M. 2003. Mammal Tracks and Sign: A Guide to North American Species. Stackpole Books, Mechanichsburg, PA. 792 pp. Fisher, J.T., and L. Wilkinson. 2005. The response of mammals to forest fire and timber harvest in the North American boreal forest. Mammalian Review 35:51–81. 692 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 Northeastern Naturalist Vol. 20, No. 4 Franklin, J.F., T.A. Spies, R.V. Van Pelt, A.B. Carey, D.A. Thornburgh, D. Rae Berg, D.B. Lindenmayer, M.E. Harmon, W.S. Keeton, D.C. Shaw, K. Bible, and J. Chen. 2002. Disturbances and structural development of natural forest ecosystems with silvicultural implications, using Douglas-fir as an example. Forest Ecology and Management 155:399–423. Fuller, A.K., D.J. Harrison, and H. J. Lachowski. 2004. Stand-scale effects of partial harvesting and clearcutting on small mammals and forest structure. Forest Ecology and Management 191:373–386. Galik, C.S., R. Abt, and Y. Wu. 2009. Forest-biomass supply in the southeastern United States: Implications for industrial roundwood and bioenergy production. Journal of Forestry 107:69–77. Getz, L.L., J.E. Hofmann, B. McGuire, and M.K. Oli. 2004. Population dynamics of the Northern Short-tailed Shrew, Blarina brevicauda: Insights from a 25-year study. Canadian Journal of Zoology 82:1679–1686. Glennon, M.J., and W.F. Porter. 2007. Impacts of land-use management on small mammals in the Adirondack Park, New York. Northeastern Naturalist 14:323–342. Greenberg, C.H. 2002. Response of White-footed Mice (Peromyscus leucopus) to coarse woody-debris and microsite use in southern Appalachian treefall gaps. Forest Ecology and Management 164:5766. Gunther, P.M., B.S. Horn, and G.D. Babb. 1983. Small-mammal populations and food selection in relation to timber harvest practices in the Western Cascade Mountains. Northwest Science 57:32–44. Hooven, E.F., and H.C. Black. 1976. Effects of some clearcutting practices on small-mammal populations in Western Oregon. Northwest Science 50:189–208. Hornbeck, J.W., and W. Kropelin. 1982. Nutrient removal and leaching from a whole-tree harvest of northern hardwoods. Journal of Environmental Quality 11:309–316. Klenner, W., and T.P. Sullivan. 2009. Partial and clearcut harvesting of dry Douglas-fir forests: Implications for small-mammal communities. Forest Ecology and Management 257:1078–1086 Levin, S.A. 1992. The problem of pattern and scale in ecology. Ecology 73:1943–1967. Litvaitis, J.A. 2001. Importance of early-successional habitat to mammals in eastern forests. Wildlife Society Bulletin 29:466–473. Loeb, S.C. 1999. Responses of small mammals to coarse woody-debris in a southeastern pine forest. Journal of Mammalogy 80:460–471. Lomolino, M.V., and D.R. Perault. 2000. Assembly and disassembly of mammal communities in a fragmented temperature rain forest. Ecology 81:1517–1532. Mahmood, S., R.D. Finlay, and S. Erland. 1999. Effects of repeated harvesting of forest residues on the ectomycorrhizal community in a Swedish spruce forest. New Phytologist 142:577–585. Mazerolle, M.J. 2006. Improving data analysis in herpetology: Using Akaike’s information criterion (AIC) to assess the strength of biological hypotheses. Amphibia-Reptilia 27:169–180. Nitterus, K., M. Astrom, and B. Gunnarsson. 2007. Commercial harvest of logging residue in clear-cuts affects the diversity and community composition of ground beetles (Coleoptera: Carabidae). Scandinavian Journal of Forest Research 22:231–240. Orrock, J.L., J.F. Pagels, W.J. McShea, and E.K. Harper. 2000. Predicting presence and abundance of a small-mammal species: The effect of scale and resolution. Ecological Applications 10:1356–1366. Northeastern Naturalist Vol. 20, No. 4 D.A. Patrick, C. Laxton, D. Ball, S. Collins, S. Korzec, C. Langevin, and J. Vimislik 2013 693 Patrick, D.A., M.L. Hunter, Jr., and A.J.K. Calhoun. 2006. Effects of experimental forestry treatments on a Maine amphibian community. Forest Ecology and Management 234:323–332. Plant, R.E. 2012. Spatial data analysis in agriculture and ecology using R. CRC Press, Boca Raton, FL. Xv + 631 pp. Riffelli, S., J. Verschuyl, D. Miller, and T.B. Wigley. 2011. Biofuels harvests, coarse woody-debris, and biodiversity—a meta-analysis. Forest Ecology and Management 261:878–887. Rittenhouse, T.A.G., E.B. Harper, L.R. Rehard, and R.D. Semlitsch. 2008. The role of microhabitats in the desiccation and survival of amphibians in recently harvested oakhickory forest. Copeia 2008:807–814. Rudolphi, J., and L. Gustafsson. 2005. Effects of forest-fuel harvesting on the amount of deadwood on clear-cuts. Scandinavian Journal of Forest Research 20:235–242. Saab, V. 1999. Importance of spatial scale to habitat use by breeding birds in riparian forests: A hierarchical analysis. Ecological Applications 9:135–151. Stevens, P.A., D.A. Norris, T.G. WIlliams, S. Hughes, D.W.H. Durrant, M.A. Anderson, N.S. Weatherley, M. Hornung, and C. Woods. 1995. Nutrient losses after clearfelling in Beddgelert Forest: A comparison of the effects of conventional and whole-tree harvest on soil water chemistry. Forestry 68:115–131. Sullivan, T.P., D.S. Sullivan, and P.M.F. Lindgren. 2000. Small mammals and stand structure in young pine, seed tree, and old-growth forest, southwest Canada. Ecological Applications 10:1367–1383. Tierney, G., and D. Faber-Langendoen. 2009. NPS northeast temperate network long-term forest monitoring protocol. National Park Service. Fort Collins, CO. Trani, M.K., R.T. Brooks, T.L. Schmidt, V.A. Rudis, and C.M. Gabbard. 2001. Patterns and trends of early-successional forests in the eastern United States. Wildlife Society Bulletin 29:413–424. Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of Wildlife Management 47:893–901. Waters, J.R., and C.J. Zabel. 1995. Northern Flying Squirrel densities in fir forests of Northeastern California. The Journal of Wildlife Management 59:858–866. Wheatley, M., J.T. Fisher, K. Larsen, J. Litke, and S. Boutin. 2005. Using GIS to relate small-mammal abundance and landscape structure at multiple spatial extents: The Northern Flying Squirrel in Alberta, Canada. Journal of Applied Ecology 42:577–586. Wintle, B.A., and D.C. Bardos. 2006. Modeling species-habitat relationships with spatially autocorrelated observation data. Ecological Applications 16:1945–1958. Zar, J.H. 1999. Biostatistical Analysis. Prentice Hall, Upper Saddle River, NJ. 662 pp.. Zielinski, W.J. 1995. Baited track-plate stations. Pp. 67–89, In W.J. Zielinski and T.E. Kucera, (Eds.). A detection manual for Wolverine, Fisher, Lynx, and Marten in the Western United States. US Department of Agriculture, Forest Service, Pacific Southwest Research Station, General Technical Report PSW-GTR 157, Albany, CA. Zwolak, R. 2009. A meta-analysis of the effects of wildlife, clearcutting, and partial harvest on the abundance of North American small mammals. Forest Ecology and Management 258: 539–545.