Trembling Aspen Stand Response 15 Years after Windthrow, Salvage Harvesting, and Forest Renewal

Abstract: Windthrow is a common occurrence in boreal forests, affecting wood supply and presenting regeneration challenges for forest managers. Salvage harvesting is often conducted afterwards to extract valuable wood and improve access for forest renewal activities. Research efforts, however, are generally limited to the effects of windthrow and salvage harvesting in the first few years following disturbance. In 2006, a catastrophic wind event occurred in a trembling aspen (Populus tremuloides Michx.) forest in northeastern Ontario. A field study was established with a range of silvicultural treatments from leaving after windthrow (W) to forest renewal treatments including windrow site preparation, planting, and herbicide release (WSPR). While the results of first 5-year assessment have been reported, the objective of the current study was to reassess treatment effects at 15 years post-disturbance, a stage of stand development that is more indicative of future forest conditions. Compared to the results of the earlier assessment, the 15-year assessment indicated that standing dead wood (snags) declined, whereas coarse downed wood did not change substantially over time. Post-disturbance salvage harvesting improved aspen regeneration in terms of density, stocking, and growth, but substantially reduced the proportion of conifers relative to windthrow only. Thus, salvage harvesting helped to sustain aspen composition, compared to other treatment options that increased proportions of pine and spruce trees from W to windthrow and salvage harvesting combined with windrow site preparation and planting (WSP), and from WSP to WSPR.

Windthrow reduces the aspen overstory releasing sub-canopy and understory trees [7,8] and pits and mounds formed by uprooted trees act as seedbeds for regenerating trees and vegetation [11,12]. Additionally, the overstory reduction is similar to that created by a clearcut harvest in that it reduces apical dominance and increases light and temperature promoting vegetative aspen regeneration [7,[13][14][15][16]. The influences of postwindthrow residual trees, standing dead trees, and downed wood on the density, stocking, and growth of aspen regeneration have been rarely quantified nor has the response of understory trees and vegetation to a catastrophic wind event.

Study Area
Field plots were established after the 2006 catastrophic windthrow, salvage harvesting, and forest renewal operations [7]. Most pre-disturbance stands were mature trembling aspen, with some aspen and black spruce mixedwoods. Typical tree responses to the 120 to 150 km h −1 winds were the snapping or twisting off of aspen stems 2 to 8 m aboveground while conifers, mostly black spruce, were uprooted. The operational salvage logging was carried out in winter in select areas of pure aspen (≥80% basal area) in 2006-2007 and 2007-2008 to recover merchantable aspen, spruce, and jack pine and to improve access for forest renewal activities. The windthrown stands salvage logged in winter 2006-2007 were site prepared by windrowing debris and planted with black spruce and white spruce. To further emulate typical Canadian boreal mixedwood renewal strategies [27], some of the planted areas were aerial sprayed with glyphosate with intent to improve spruce survival and growth.

Experimental Design
The 15-year post-windthrow assessments, conducted in the same measurement plots, established the following treatments in the complete randomized design described in detail by Man et al. [7]. The

Standing and Downed Wood Surveys
In each of the 25 experimental units, live trees and standing dead stems (snags) were assessed in 3 circular overstory plots of 11.28 m radius, with at least 40 m between plot centres. The original overstory survey included trees ≥ 4 m in height [7]. The year 15 re-assessment added an additional threshold of diameter at breast height (DBH) ≥ 10 cm to exclude post-disturbance regeneration that was captured in the regeneration assessments. All live overstory trees and standing snags (≥0.5 m in height) were tallied for species, DBH, and height. Some snags were decayed such that species identification was limited to broadleaf or conifer. Honer et al.'s equations [30] for central and eastern Canada were used to calculate the volume of overstory trees, while snag volume was calculated as a cylinder with diameter (diameter at breast height) and length (snag height), as suggested by Man et al. [7].
Coarse downed wood was surveyed along three 11.28 m transects emanating from the overstory plot centres. The azimuth of the first transect was randomly selected with the second and third transects at increments of 120 • . Using the criteria outlined in Hayden et al. [31], coarse wood species, decay class, and diameter at the transect intersection midpoint were recorded for each piece ≥ 7.5 cm in diameter. Down coarse wood in decay classes 4-5 was often difficult to identify as a certain species or even as a species group, leaving some pieces unidentified. Coarse wood volume was determined using the method described by Van Wagner [32].

Regeneration and Vegetation Assessments
A line-plot method was used to assess regeneration and vegetation in July-August 15 years post-windthrow [7]. Lines were parallel to the long axis of the survey area or perpendicular to windrowed slash piles in treatments WSP and WSPR. All experimental units had 27 circular survey plots, established along a minimum of 3 parallel survey lines more than 20 m apart at 10 m intervals. Each 4 m 2 survey plot was used to measure tree species, height, and DBH of regenerating seedlings and saplings (height ≥ 2 cm and DBH < 10 cm), and assess the ground cover of downed wood and vegetation. Percent cover was recorded for coarse downed wood (≥7.5 cm in diameter for downed woody material and stumps <2 m height), fine downed wood (<7.5 cm in diameter, mainly small trees and fine branches), shrubs, herbs, ferns and allies, grasses and sedges, mosses, and lichens. Regeneration stocking was calculated as the percentage of regeneration plots with at least 1 crop tree.

Data Analysis
Data analysis followed a completely randomized design with 5 replications. The analysis of variance (ANOVA) linear mixed effects (LME) model, available in nlme package, was used to compare differences in overstory live tree volume and density, snag volume and density, coarse downed wood volume, percent cover of grouped vegetation and coarse and fine downed wood, and regeneration density, stocking, height, and diameter among 5 post-windthrow treatments. Data were transformed when necessary to achieve normality based on a visual assessment and Shapiro-Wilk's test of residuals. Multiple contrasts of means among treatments were performed using Tukey HSD test with p values adjusted with Shaffer method available in multcomp package. All analyses were implemented using R version 4.1.2 [33].

Overstory Trees and Snags
The volume of overstory trees ranged from 2.5 m 3 ha −1 in WSPR to 15.2 m 3 ha −1 in W, and overstory density ranged from 70 stems ha −1 in WSP to 185 stems ha −1 in W (Figure 1a,b). While both overstory volume and density did not differ significantly among the treatments (p = 0.382 and 0.253, respectively), overstory trees were mostly trembling aspen in WS and C, spruce trees in WSPR, jack pine in WSP, and balsam fir in W.

Downed Wood
The volume of coarse downed wood was higher in W and WS treatments than in areas with renewal operations (WSP and WSPR) and the clearcut (Figure 1e, p < 0.001). The most frequent decay class of the coarse downed wood was 3, with few in class 1 or 5 ( Figure 1f). The proportion of ground area covered by coarse downed wood showed a similar pattern, but was reduced by herbicide release in WSPR treatment relative to WSP (Figure 2a, p < 0.001).
Comparatively, the area covered by downed fine wood differed significantly among the five treatments ( Figure 2b, p < 0.001). All salvage harvesting and renewal operations reduced fine wood, with the lowest cover in the WSPR treatment. The volume of standing snags ranged from 4.8 m 3 ha −1 in WSPR to 35.3 m 3 ha −1 in W (Figure 1c, p = 0.135) treatment. Standing snags were mostly trembling aspen (60%) and white birch (20%) and numbers were marginally higher in W (97 pieces ha −1 ) than in WSPR and C (about 20 pieces ha −1 ) (Figure 1d, p = 0.059) treatments. The difference between treatments W and WS can be attributed to conifer removal during salvage harvesting operations.

Downed Wood
The volume of coarse downed wood was higher in W and WS treatments than in areas with renewal operations (WSP and WSPR) and the clearcut (Figure 1e, p < 0.001). The most frequent decay class of the coarse downed wood was 3, with few in class 1 or 5 (Figure 1f). The proportion of ground area covered by coarse downed wood showed a similar pattern, but was reduced by herbicide release in WSPR treatment relative to WSP (Figure 2a, p < 0.001).

Vegetation Cover
Shrub cover was lower (p < 0.001) and grass and sedge cover was higher (p = 0.031) in WSPR than in the other 4 treatments, which did not differ significantly from one another (Figure 3a,c). The total cover of herbaceous species was highest in the WSPR treatment, followed in decreasing order by C, WSP, W, and WS treatments (Figure 3b Comparatively, the area covered by downed fine wood differed significantly among the five treatments (Figure 2b, p < 0.001). All salvage harvesting and renewal operations reduced fine wood, with the lowest cover in the WSPR treatment.

Vegetation Cover
Shrub cover was lower (p < 0.001) and grass and sedge cover was higher (p = 0.031) in WSPR than in the other 4 treatments, which did not differ significantly from one another (Figure 3a,c). The total cover of herbaceous species was highest in the WSPR treatment, followed in decreasing order by C, WSP, W, and WS treatments (Figure 3b

Tree Regeneration
Among the four post-windthrow treatments, broadleaf regeneration density and stocking were higher in WS and WSP than in W and WSPR, while broadleaf regeneration height and diameter did not differ between W and WSP (Figure 4a,c,e,g, p < 0.001). The C treatment had the highest broadleaf tree density, stocking, height, and diameter, but the difference with WS was significant only for total height ([ 4e). Most broadleaf regeneration was trembling aspen, except in treatment WSPR where glyphosate killed the aspen and left less susceptible white birch (Figure 4a).
Conifer regeneration density and stocking were highest in the W treatment, followed by WSP or WSPR (Figure 4b,d; p < 0.001). Regeneration in WS and C was negligible. Among the regenerating conifers, WSPR was almost solely spruce (99%); WSP was mixed spruce (49%), pine (24%), and fir (27%); and W was mostly fir (81%) with some spruce (19%). The regenerating conifers in the WSPR treatment were 3.7 m in height and 5.3 cm in diameter, with the latter significantly larger than that in the other treatments ( Figure  4h, p = 0.005).

Tree Regeneration
Among the four post-windthrow treatments, broadleaf regeneration density and stocking were higher in WS and WSP than in W and WSPR, while broadleaf regeneration height and diameter did not differ between W and WSP (Figure 4a,c,e,g, p < 0.001). The C treatment had the highest broadleaf tree density, stocking, height, and diameter, but the difference with WS was significant only for total height (Figure 4e). Most broadleaf regeneration was trembling aspen, except in treatment WSPR where glyphosate killed the aspen and left less susceptible white birch (Figure 4a).

Post-Disturbance Dynamics
As with wind disturbances in other forests [9,34], the 2006 windthrow removed much of the aspen stand overstory, creating large quantities of snags and downed dead wood. The year 15 post-disturbance assessment showed an apparent recovery, with overstory volume in W nearly doubled and density nearly quadrupled compared to earlier assessments [7]. The overstory change in W can be attributed to the release of understory balsam Conifer regeneration density and stocking were highest in the W treatment, followed by WSP or WSPR (Figure 4b,d; p < 0.001). Regeneration in WS and C was negligible. Among the regenerating conifers, WSPR was almost solely spruce (99%); WSP was mixed spruce (49%), pine (24%), and fir (27%); and W was mostly fir (81%) with some spruce (19%). The regenerating conifers in the WSPR treatment were 3.7 m in height and 5.3 cm in diameter, with the latter significantly larger than that in the other treatments (Figure 4h, p = 0.005).

Post-Disturbance Dynamics
As with wind disturbances in other forests [9,34], the 2006 windthrow removed much of the aspen stand overstory, creating large quantities of snags and downed dead wood. The year 15 post-disturbance assessment showed an apparent recovery, with overstory volume in W nearly doubled and density nearly quadrupled compared to earlier assessments [7]. The overstory change in W can be attributed to the release of understory balsam fir and spruce [7,8]. Due to the ingress of released understory trees and dieback of old residual trees [27,35] from wind and decadence [36], the average size of overstory trees decreased for both height and diameter.
At the same time, the volume of standing snags in the W treatment decreased by >40% and density by >50%. The larger reduction in density suggests a greater loss of small-diameter snags over time [6,37], due to higher rates of decay [38] and susceptibility to physical movement by wind or wildlife. Average snag diameter increased from 24.5 cm shortly after disturbance to 25.8 cm at year 15. In the W treatment, the decrease in standing snags at year 15 resulted in additional inputs to coarse downed wood volume and percent cover, which remained similar to those observed shortly after disturbance [7].
Comparatively, the cover of fine downed wood decreased substantially, likely due to lack of input from live trees and higher rates of decay for small downed wood [38].
Despite the residual overstory, snags, and downed wood that could interfere with the abundance and growth of the aspen suckers [13], windthrow stimulated aspen vegetative regeneration, consistent with suggestions that windthrow events benefit the regeneration of pioneer species [39][40][41]. Meanwhile, the windthrown conifers create pits and mounds with a gradient of soil and microclimates [11,12], conditions that encourage the germination and natural regeneration of balsam fir and spruce [7]. Compared to broadleaf regeneration density, which decreased through self-thinning over time [42], conifer density increased in the W treatment, due to natural recruitment of shade tolerant balsam fir in both overstory and understory [7].

Salvage Logging
Consistent with Waldron et al. [4] and Taylor et al. [9], salvage harvesting operations reduced the abundance of overstory trees, snags, and downed wood at year 15 postdisturbance. Yet, over time, the effects of salvage harvesting weakened and generally became insignificant, likely due to the low harvesting intensity (30%) relative to typical operational levels [24]. The level of salvage would have been higher if market demand for aspen wood had been stronger [7].
The 15-year post-windthrow assessment continues to support the use of operational salvage harvesting for aspen regeneration in terms of resulting density, stocking, and growth, as expected given aspen regeneration biology [13]. These results are consistent with those after prompt post-fire salvage operations in aspen stands [14,15] but differ from those after delayed salvage harvesting [16]. Salvage operations that immediately follow windthrow allow the benefits of the salvage removal to exceed the mechanical damage to aspen regeneration; a delay shifts the benefit-damage balance. Additionally, it was observed in this study [7] and others [22,23] that the salvage operation could damage advanced regeneration of shade tolerant species, such as spruce and balsam fir, although logging operations also add seedbeds for natural regeneration of these species [20][21][22]. As a result, salvage harvesting may not change total regeneration density and stocking [24] but may increase the proportion of early successional species, i.e., trembling aspen, jack pine, and white birch [9,20,34,43].
Other than a trend of lower lichen cover, salvage harvesting had minimal effects on vegetation abundance [7,44], probably due to the relatively low harvest removal [7], the limited forest floor disturbance by the winter operation [45], and a quick increase in broadleaf regeneration and shrub cover throughout the treatment areas [7]. The lower lichen cover in salvage logged areas [46,47] is probably due to lichen sensitivity to light and moisture conditions and slow recovery following disturbances [35,42,48].

Forest Renewal Operations
To ensure adequate spruce or pine composition in post-disturbance stands, forest renewal operations are required to increase the density and stocking of the desired crops [49,50]. As per early observations, windrowing coarse wood and planting conifers did not significantly affect the remaining overstory and broadleaf regeneration [7]. However, the windrow site preparation knocked down standing snags and reduced the land area covered by coarse downed wood. Additionally, the windrowing process moved cones across the site, exposed forest floor for seedbeds, and increased light and temperature, all helping the germination of natural jack pine, spruce, and balsam fir [51]. The planted and naturally regenerated trees were adequate in density and stocking, but substantially smaller in diameter than what would be in the Windthrow-Salvage-Plant-Release treatment, which benefits from both planting and vegetation control [49,50].
The reduction in coarse downed wood volume at year 15 in windrowed treatments WSP and WSPR relative to WS is unexpected and differs from early observations [7]. The change is likely due to further decomposition and compaction of coarse wood windrows, adding to the difficulty in identifying pieces underneath wood piles and leading to possible underestimation.
The herbicide release with glyphosate substantially reduced tall vegetation (i.e., overstory, understory broadleaves, and shrubs) and increased the cover of lower vegetation (i.e., herbs and grasses) that responds quickly to disturbance [52,53]. White birch is less responsive to foliar spraying with glyphosate than trembling aspen [54] and it became the dominant broadleaf regeneration in the WSPR treatment. Even so, white birch growth was substantially reduced relative to trees in non-sprayed treatments [49] and not adequate to overtop the conifers. The year 15 assessment confirmed early results from this study [7] and other studies [49,50,55] that reducing tall vegetation benefits the survival and growth of natural and planted conifers.

Management Considerations and Conclusions
The early effects of windthrow, salvage, windrow site preparation, planting, and herbicide release on overstory trees, standing and down coarse wood, vegetation cover, and regeneration persisted 15 years post-disturbance. The catastrophic windthrow removed much of the overstory, releasing sub-canopy and understory trees and creating large amounts of standing and down coarse wood. Relative to the C treatment, aspen regeneration on other treatment sites was impeded by the coarse wood; however, the abundance and growth of the aspen regeneration in the W, WS, and WSP treatments will be adequate for aspen dominance in future forests. The windthrow event will accelerate the transition of aspen stands (≥80%) to aspen-dominated mixedwoods (50%-80%), similar to the dynamics of aspen stands after defoliation by forest tent caterpillar and the resulting decline [56].
The salvage harvesting only treatment (i.e., WS) removed some of the residual trees and snags, which helped aspen regeneration in terms of density, stocking, and vegetative growth. However, the harvesting operation reduced advanced growth of spruce and balsam fir. If the forest management objective is to restore wind-affected stands to pre-disturbance conditions, a salvage harvest shortly after windthrow is a viable option, and may also benefit the natural regeneration of jack pine and spruce. With the salvage harvesting prescription, the year 15 overstory and understory composition suggests a future forest of pure aspen.
The use of planting and release (i.e., treatments WSP and WSPR) are common practices for increasing conifers in aspen stands after clearcut. The WSP treatment increased conifer abundance but growth of planted spruce was lower than that in the WSPR treatment, due to competition from abundant aspen regeneration and shrubs. The WSP treatment indicates future stands of aspen-dominated mixedwoods similar to those in the untreated windthrow area. If the management objective is a conifer-dominant stand, a post-planting release treatment is needed, as is common practice in Canadian boreal forests following a clearcut [27,35,49,50].
Author Contributions: R.M. and M.R. contributed equally to study design, data collection, interpretations of results, and writing. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.