Draft Spatial partitioning of competitive effects from neighbouring herbaceous vegetation on establishing hybrid poplars in plantations

Journal: Canadian Journal of Forest Research Manuscript ID cjfr-2018-0410.R1 Manuscript Type: Article Date Submitted by the Author: 17-Dec-2018 Complete List of Authors: Goehing, Jeannine; University of Alberta, Renewable Resources Henkel-Johnson, David; University of Alberta, Renewable Resources Macdonald, S. Ellen; University of Alberta, Bork, Edward; Department of Agriculture, Food and Thomas, Barb; University of Alberta Keyword: aboveground and belowground competition, resource availability, vegetation management, Hybrid poplar, tree growth Is the invited manuscript for consideration in a Special Issue? : Not applicable (regular submission)


Introduction
Plantations of fast-growing hybrid poplar trees play a significant role in wood and fibre production, and as a supplier of biomass for energy, alternative fodder sources, and ecosystem services such as carbon sequestration (Weih 2004, Poplar Council of Canada 2012).In the Canadian prairie provinces, plantations are established on land formerly in conventional agricultural production (e.g., annual crops or forage).Key characteristics of these systems include a reduced rotation length from 60-80 years for aspen to less than 20 years for hybrid poplar, and intensive silvicultural management, which includes a combination of chemical and mechanical treatments to control competing herbaceous vegetation (Hansen 1983, Weih 2004).
Weed control in plantations aims to reduce tree yield loss and mortality through a reduction in resource competition, either aboveground for light or belowground for nutrients and/or water (Balandier et al. 2006) and direct physical competition (Henkel-Johnson et al. 2016).Hybrid poplars are known to have high requirements for light, water and nutrients, particularly nitrogen (N), and do not reach their full growth potential when experiencing climate-, site-or competition-induced limitations.
Moreover, boreal soils often demonstrate nutrient limitations for tree growth.Several studies report plantation productivity to be predominantly limited by belowground competition for water and/or nutrients, rather than by aboveground competition for light (Coll et al. 2007, Pinno andBélanger 2009).
However, Sixto et al. (2001) found poplars were also intolerant of shading by neighboring herbaceous vegetation during the early establishment period.
Early vegetation control is critical to maintaining rapid initial growth of hybrid poplar plantations (Otto et al. 2010) and reducing tree yield losses later in the growth cycle (Truax et al. 2012).Therefore, vegetation management typically focuses on controlling vegetation during the establishment phase when poplar trees are most sensitive to competition and have high demand for resources (Hansen et Stanturf et al. 2001, West 2006).During this phase, neighboring vegetation has complex competitive and facilitative effects on poplar trees, which may be influenced by its abundance and growth form (Grenke et al. 2016).In addition to temporal and compositional variation in the effects of competing vegetation, the impact of adjacent herbaceous vegetation on tree performance varies spatially.Most studies report greater effects of competition from neighboring vegetation close to trees (e.g., 0.5-1 m) and emphasize the need to control this vegetation (Thomas et al. 2001, Powell andBork 2004).However, the widely used mechanical vegetation control methods (cultivation or mowing) can result in equipment-induced damage to branches, stems, and roots of trees; thus it is impossible to safely treat areas very close to the tree base.Little is known about how the spatial relationships among trees and neighboring vegetation influence competitive effects, nor how this varies with plantation age.
The objective of this study was to determine the spatial effects of competition with herbaceous vegetation on the performance of hybrid poplar, including comparison of aboveground-versus belowground-competition, as well as competition near and far from the tree stem.To achieve this we assessed performance of two hybrid poplar clones over a four year period in response to six vegetation control treatments.Specifically, we tested: (1) the relative importance of herbaceous competition near versus far from the stem, and above-versus belowground, on the growth of two hybrid poplar clones; and (2) the effect of the vegetation control treatments on neighboring herbaceous vegetation, light, soil nutrients, moisture, and temperature, and linked these environmental conditions to the observed growth responses of poplar trees.

Study area and Experimental design
The study was conducted at an experimental plantation in central Alberta, Canada near Boyle (54°90′N, 112°85′W, 570 m), in the Dry Mixedwood Natural Subregion (Natural Regions Committee 2006).The climate is temperate continental and characterized by short, warm growing seasons with 30 yr average (1981 to 2010) mean July temperatures of 16.6 °C, and long, cold winters with a mean January temperature of -13.4 °C.The 30 yr normal annual precipitation is 479 mm, of which 336 mm (i.e., 70%) falls during the growing season from May through September (Environment Canada 2014).
Herbaceous vegetation on all sites was dominated by a diverse mix of introduced weedy forbs (annuals and perennials) and graminoids (Supplementary Information Table S1).
The study was established in spring 2011 on research sites managed by Alberta-Pacific Forest Industries Inc. (Al-Pac), and tree-weed competition dynamics were monitored over a four-year period through fall 2014.The experimental design was a split-split-plot at each of three sites (whole plot), with two hybrid poplar clones (split-plot) and six vegetation control treatments (split-split plot) included in each of six blocks per site (Supplementary Information Fig. S1).All sites were located in a larger fenced area to protect trees from browsing by ungulates such as deer and moose.The three study sites were placed in different topographic locations, representing the range on which poplar plantations might be established in the region: a low-lying area characterized by imperfect drainage (Lowland), a mesic site (Midland), and a rapidly drained upland site with a west-facing aspect (Highland).Soils were characterized as Orthic Gray Luvisols (Alberta Soil Information Viewer 2014).
The two hybrid poplar clones were the female clone Walker (Populus deltoides x (P.laurifolia x P. nigra)) (Lindquist et al. 1977) and male clone Okanese (Walker x (P.laurifolia x P. nigra)) (Schroeder et al. 2013).The two related intersectional hybrids were selected due to their economic importance in D r a f t 6 shelterbelts, and more recently, in short-rotation-intensive-culture (SRIC) plantations across the Canadian Prairies, and their contrasting growth forms, resource requirements, and sensitivity to competition (van Oosten 2004, 2006, Schroeder et al. 2013).
At each site we established six blocks (28 m X 11.2 m); these were each divided into two splitplots that were randomly assigned to be planted to either the Walker or Okanese clone (Supplementary Information Fig. S1).Trees were grown from cuttings (10 cm) at a commercial nursery (Smoky Lake Forest Nursery near Smoky Lake, Alberta) in summer 2010, then packaged in fall 2010 and stored at -2.5 °C until planting in July 2011 (Dave Kamelchuk, Al-Pac, personal communication, 2014).Mean root-collar diameter / height at planting were: Okanese 3.48 mm / 34.23 cm, Walker 2.59 mm / 23.33 cm.Trees were hand-planted in each split-plot at a 2.8 m grid spacing in a 5 × 5 configuration, which included an outside buffer row of trees that were not sampled (Supplementary Information Fig. S1).Trees were not fertilized at any time during the study.From the nine trees in the interior portion of each split-plot (i.e., 3 x 3 tree grid), the six most uniform trees were selected and each was randomly assigned to one of six vegetation control treatments (split-split plot): (1) Control: No removal of vegetation (NR; no removal); (2) Removal of aboveground vegetation close (0-50 cm) to the tree stem (AC; above-close); (3) Removal of aboveground vegetation far (50-140 cm) from the tree stem (AF; above-far); (4) Removal of aboveground vegetation from 0-140 cm (AT; above-total); (5) Removal of above-and belowground vegetation close (0-50 cm) to the tree stem (BC; above & below-close); and (6) Removal of above-and belowground vegetation far (50-140 cm) from the tree stem (BF; above & below-far).

Application of treatments
The vegetation control treatments were applied within a 140 cm radius around each tree stem; this was further divided into the zones near the tree (0-50 cm) and far from the tree (50-140 cm).D r a f t 7 season using a hand-held weed whacker to trim neighboring vegetation to ground level.Resulting litter was left evenly distributed across the treatment area.For the removal of both above-and belowground vegetation, hand application of a 10% glyphosate solution was used in combination with the installation of plastic root exclusion barriers.Herbicide was applied during the early morning using hand spray bottles while shielding trees to prevent drift.For the above&below-close and above&below-far treatments, we installed circular root exclusion barriers (150 µm thick clear plastic, to a depth of 15 cm) in summer 2011 at a distance of 50 cm from the base of each tree stem.These barriers prevented root incursion from adjacent uncontrolled vegetation into the area subjected to removal of above-and belowground vegetation.Vegetation control started at the time of planting in July 2011 for all treatments and was repeated periodically (e.g., monthly, or as needed), throughout the growing seasons of 2011, 2012, and 2013, but not 2014.

Data collection
Initial height and basal root collar diameter were measured on all trees in July 2011 shortly after planting.To facilitate repeated measures of diameter, stems were permanently marked 3 cm above the ground.Total tree height and diameter were measured again for all live experimental trees between late September and late October of 2011, 2012, 2013, and 2014 (the end of the first to fourth growing seasons, respectively).Height was measured from ground level on straightened trees to the base of the terminal bud.Root collar diameter was measured with a digital caliper in two perpendicular directions (N-S and W-E) at the marked portions of stems, which were then averaged prior to analysis.For the 2011 growing season, height and diameter growth increments were calculated as the difference between initial (at planting) and fall measurements; for 2012, 2013, and 2014, increments were calculated as the difference between fall measurements of two consecutive years.Tree survival was recorded at the end of each growing season.Survival was calculated for each clone and treatment D r a f t 8 combination and expressed as a percentage of the original total number of living trees across all 18 blocks.No statistical analyses were undertaken for the survival data; dead trees were excluded from analysis of growth increments.
We measured the degree of cover of herbaceous vegetation around all living experimental trees in August of the second growing season (2012).Sampling was done out to a 100 cm distance from each tree in all four cardinal directions using belt transects comprised of quadrats (25 cm wide x 50 cm long) designed to sample vegetation cover at two distances (near: 0-50 cm; far: 50-100 cm) from the tree base (Supplementary Information Fig. S1).Within each quadrat, cover was visually estimated for each vascular plant species (within 5% up to 20%, within 10% above 20%).Plant species were also grouped based on growth form and life cycle, including: 1) annual forbs (including winter annuals and biennials), 2) perennial forbs, 3) annual grasses, and 4) perennial grasses (Supplementary Information Table S1).
Herbaceous vegetation cover around each tree, summed for all species and by functional groups, was calculated for: 1) the area 0-100 cm from the tree (cover values were averaged for all eight quadrats per tree); and 2) for the area close to (0-50 cm) and farther from (50-100 cm) the tree (cover values averaged for the four quadrats at one or the other sampling distance).
Soil nutrient availability was quantified using Plant Root Simulator (PRS) probes containing ion exchange resin membranes (Western Ag Innovations, Inc., Saskatoon, Canada) that were installed in the second and third growing seasons (2012 and 2013, respectively).Four (out of six) blocks were randomly selected at each site and probes were installed adjacent to the trees subjected to the following vegetation control treatments: control, above-close, above-far and above-total.Due to budgetary constraints we were unable to assess nutrient availability in all six vegetation control treatments; our assumption was that the herbicide treatments that removed both above-and belowground vegetation would have had similar, albeit more dramatic, effects as compared to the removal of aboveground vegetation.PRS-probes (each pair containing one anion and one cation probe) were vertically inserted D r a f t 9 into the soil at two distances from the stem (25 and 95 cm) for a total of 192 PRS-samples (3 sites × 4 blocks × 4 treatments × 2 clones × 2 distances).At each distance, one pair of PRS probes was buried in each cardinal direction for a total of four pairs per distance, and eight pairs per tree in total.Probes were installed approximately 12 cm deep before the first vegetation control treatment was applied during the growing season, and left in place for 10 weeks in 2012 (June 18 to August 23) and nine weeks in 2013 (May 21-24 to July 20).After removal, all probe pairs for a given distance around a given tree were combined for analysis.All probes were cleaned with deionized water and shipped to Western Ag Innovations Inc. and analyzed for NO 3 -, NH 4 + , PO 4 3− , K + , SO 4 2-, Ca 2+ , Mg 2+ , Mn 2+ , Al 3+ , Fe 2+ , Cu 2+ , Zn 2+ , B + , Pb 2+ , and Cd 2+ .PRS-probe supply rates are reported as μg of nutrient/10 cm 2 /burial period.
Volumetric soil moisture content (%) to 5 cm depth was measured once (August 18) in 2012 and three times (June 5, July 4, August 17) in 2013, with a ML2x ThetaProbe soil moisture sensor attached to a HH2 moisture meter (Delta-T Devices, Cambridge, UK), at least two days after significant precipitation.Three measurements were taken at each of two distances (25 cm and 95 cm) in random directions from each experimental tree.The three measurements at each distance were averaged prior to analysis.Peak soil temperature (°C) was measured once (July 24) in 2012, and three times (June 2, July 24, August 17) in 2013, between 14:00 and 16:00 MDT, using a 450ATT digital soil thermocouple thermometer (Omega, Laval, PQ, Canada), following the same sampling design and distances as for soil moisture.
Photosynthetically active radiation (PAR; 400-700 nm) was measured during a two-hour period around solar noon during stable weather conditions (either clear sky or completely overcast) on June 11 and July 4, 2013, using an 80 cm long sunfleck ceptometer (AccuPAR, Decagon devices, Inc., Pullman, USA).PAR was recorded: 1) above the tree and other vegetation for an unobstructed sky view; and 2) outside the tree canopy but within the competing vegetation at the vertical midpoint of the shaded portion of the tree crown (to quantify PAR transmission through the understory vegetation to the affected portion of the tree canopy).Four instantaneous readings were taken at each location for each D r a f t 10 tree and then averaged for the given height.To correct for variability in open sky PAR, the PAR measured within the competing vegetation was expressed as a percentage of PAR measured above the plant canopy for each tree individually.
To provide general information on soil conditions at each of the three sites we collected ten soil cores from random locations across the six blocks per site in August 2012.Each core was separated into depths of 0-15 cm and 15-30 cm, and then composited into one sample per depth and site and stored frozen.The soils were analyzed for texture (sand >50 μm, clay <2 μm, and silt 2-50 μm, %, hydrometer method), pH, electrical conductivity (EC, μs/cm, pH conductivity meter), organic matter (%, loss on ignition), total nitrogen (%), ammonium and nitrate (mg/kg air dried soil, colorimetrically on a SmartChem Discrete Wet Chemistry Analyzer) by the Natural Resources Analytical Laboratory at the University of Alberta, Edmonton, Canada (Table S2).

Statistical analysis
To test for treatment effects on the different response variables we used mixed-model analyses of variance (ANOVA), appropriate to the split-split-plot experimental design, using proc MIXED in SAS vers 9.2 (SAS Institute Inc. 2011).The models included site (whole-plot), type of clone (Okanese, Walker; split-plot), vegetation control treatment (NR, AF, AC, AT, BF, BC; split-split-plot), and their interactions as fixed effects; block (six per site) nested in site, and the clone × block interaction were included as random effects.We conducted preliminary analyses including year as a fixed effect following a repeated measures design; these almost always showed significant interactions between year and the other fixed effects, and thus we subsequently conducted analyses for each year separately.Site was included as a fixed factor to account for variation among the three different topographic positions (Lowland, Midland, Highland), and test for potential interactions between site and the treatments.However, there was no true replication of site types and we therefore do not make inferences about the effects of topographic D r a f t 11 position or site type.In analyses examining treatment effects on tree basal diameter and height increment we included initial tree basal diameter and height (June 2011) as covariates, when significant, to account for variation in tree size at planting.We examined treatment effects on: 1) cover of herbaceous vegetation (total and for each functional group) at a distance of 0-100cm from each tree (average of all eight quadrats per tree); 2) relative transmittance of PAR through the understory vegetation; and 3) volumetric soil moisture, soil temperature, and soil nutrient supply rates (total nitrogen, NO 3 --N, NH 4 + -N, PO 4 3− , K + , SO 4 2-, Ca 2+ , Mg 2+ , Mn 2+ , Fe 2+ , and Zn 2+ ) averaged for the measurements at 25 cm and 95 cm from the tree.We subsequently analyzed treatment effects separately for each of the two distances from the tree for: vegetation cover (0-50 cm, 50-100 cm) and soil moisture, temperature and nutrient supply rates (25 cm, 95 cm).
We tested for the assumptions of normality and equal variances using plots of residuals, and transformed the data when necessary.Significant fixed effects were further explored using lsmeans with a Bonferroni correction to family-wise α = 0.05.For example, when the vegetation removal treatment was significant we made 15 pairwise comparisons among the six treatments at α adj = 0.05/15= 0.003.
When there were significant interactions of vegetation treatment and clone we made the pairwise comparisons among vegetation treatments for each clone separately.For analyses of herbaceous vegetation, soil moisture, temperature, and nutrient supply rates at different distances from the tree, we made pairwise comparisons among the subset of vegetation control treatments that were relevant for that distance; e.g., for herbaceous vegetation 0-50 cm from the tree we made six pairwise comparisons among these four treatments: 1) control (NR); 2) above-close (AC); 3) above&below-close (BC); and 4) above-total (AT).

Hybrid poplar performance
Both clones had 100% survival through the first growing season and survival of Okanese poplar remained high (> 78%) throughout the four year study (Fig. 1).In contrast, survival of Walker poplar was more variable among treatments, dropping noticeably for the above&below-close removal treatments in year three, and for both above&below treatments and the control (no removal) in year four (survival data not subjected to statistical analysis).
The vegetation control treatments had a significant influence on tree height and diameter increment in all years (except height increment in the fourth year) (Table 1).There were significant interactions between clone and vegetation control treatments for height and diameter increment in most years (Table 1).Thus, we examined the responses to vegetation control treatments for each clone separately.Although there were significant effects of site on diameter and height increment, there were no significant interactions between site and the vegetation control treatments; thus we concluded that the effects of control treatments on diameter and height increment were similar at all sites (Table 1).
Initial tree diameter was a significant co-variate for diameter growth increment in 2011 and 2014, but not in 2012 or 2013 (Table 1).There were significant effects of treatment on diameter growth as early as the first growing season, even after adjusting for initial diameter (Table 1).In the first growing season (2011), the greatest diameter increment for both Walker and Okanese clones was for trees in the treatments where aboveground vegetation was controlled close to trees (above-close, above-total, above&below-close), while trees grew more slowly in the treatments with no vegetation control or where vegetation removal occurred farther than 50 cm from the tree (above-far, above&below-far) (Fig. 2).For the Walker clone there were no significant differences among the vegetation control treatments in terms of diameter increment in the second, third or fourth growing seasons (Fig. 2).For Okanese the lowest diameter growth increment during the second, third and fourth D r a f t 13 growing season was in the control treatment (Fig. 2).In the second growing season (2012), removal of above-and belowground vegetation close to trees resulted in the greatest diameter increment, followed by removal of above-and belowground vegetation far from trees, removal of aboveground vegetation 0-140 cm from trees, removal of aboveground vegetation close to trees, and then the removal of aboveground vegetation far from trees.Effects of treatments followed a similar pattern in the third (2013) growing season except that trees subject to the removal of above-and belowground vegetation far from the tree had a higher diameter increment than any other treatment (Fig. 2).During the fourth growing season the greatest diameter increment was for treatments that had aboveground vegetation removed close to, or from 0-140 cm from the tree (AC or AT), and above-and belowground vegetation removal far from trees (BF) (Fig. 2).Initial tree heights from the spring measurements in 2011 were significant (p< 0.05) covariates only for the first two growing seasons after planting (2011 and 2012) (Table 1).There were significant effects of the vegetation control treatments on height increment in the first three growing seasons, and significant clone by treatment interactions in the first, second and fourth growing seasons (Table 1).
Height increment of Okanese trees did not differ among vegetation control treatments during the first growing season (2011) (Fig. 2).In subsequent years, the greatest height growth occurred in plots with control of above-and belowground vegetation far from trees (BF), followed by the control of aboveground vegetation close to trees, or from 0-140 cm distance from trees (AC or AT) (Fig. 2).There were also significant effects of the vegetation control treatments on the height increment of Walker poplar, but only in the first and third growing seasons (Fig. 2); in both years the greatest height increment occurred for trees subject to the removal of aboveground vegetation from 0-140 cm (AT) (Fig. 2).

Treatment effects on herbaceous vegetation and environmental conditions
We encountered 37 different herbaceous plant species surrounding the hybrid poplar trees in this study (Supplementary Information Table S1).Adjacent herbaceous vegetation was strongly affected by control treatments (Tables 2, 3, 4).Total herbaceous cover close to trees (0-50 cm) was significantly lower in the treatment using herbicide to remove both above-and belowground vegetation (BC) compared to the other treatments that only controlled aboveground vegetation close to the tree (Fig. 3A).Notably, herbaceous cover in the above&below-close (BC) treatment was reduced by 65 -75% as compared to the treatments involving mechanical removal of aboveground vegetation (above-close, above-total) and the control treatment without vegetation suppression (Fig. 3).Total vegetation cover far from trees (50-100 cm) was also significantly lower (reduced by ~70%) in the treatment using herbicide to remove above and belowground vegetation at this distance from the tree (above&belowfar) as compared to the treatments with mechanical removal of aboveground vegetation (above-far, above-total) or the control (NR) treatment (Fig. 3).
Composition of surrounding vegetation also differed among the treatments at the end of the second growing season, primarily reflecting treatment effects on perennial plant species (Tables 3 and 4, Fig. 3).For example, the relative proportions of plant functional groups differed markedly between those treatments using herbicides to remove both above-and belowground vegetation (above&belowclose, above&below-far), as compared to all other treatments (Tables 3 and 4, Fig. 3).Both the above&below-close (BC) and above&below-far (BF) treatments resulted in lower proportional cover of perennial grasses compared to all other treatments (Tables 3 and 4, Fig. 3).Interestingly, treatments that used mechanical removal of aboveground vegetation (above-close, above-far, above-total) generally did not differ from the control treatment in terms of the cover of different plant functional groups (Tables 3 and 4), though mechanical removal did favor neighboring vegetation comprised of annual forbs rather than perennial forbs (Fig. 3).D r a f t 15 Overall, transmittance of available PAR through the understory vegetation to the shaded portion of the tree was high (ranged from 71% to 99%) but nevertheless differed among the vegetation control treatments in both June and July of the third growing season (2013) (Table 2).PAR transmission early in the growing season (i.e., June) was higher with removal of both above-and belowground vegetation (i.e., above&below-close (BC) treatment) than in all other treatments (above-close, above-total, no removal), which in turn, did not differ from one another (Table 3).Later in the season (July), however, light transmission increased with more intensive vegetation removal treatments, being lowest in the control treatment followed by above-close, then above-total, then above&below-close (Table 3).
All belowground environmental attributes measured (soil temperature, moisture and nutrient supply rates) were strongly affected by the vegetation control treatments (Table 2).For a summary of mean soil characteristics for each site and soil depth, see Supplementary Information Table S2.Soil temperature was found to differ among vegetation control treatments during all sampling times (Tables 3 and 4).Vegetation removal increased soil temperatures (by an average of 2°C) compared to zones without vegetation removal; soil temperatures were always lowest in the control treatment and greatest for treatments that removed both above-and belowground vegetation (above&below-close, above&below-far; Tables 3 and 4).
Soil moisture differed among vegetation control treatments at all sampling times for measures near the tree (25 cm from trees) (Table 3) and during August 2012 and July 2013 for measures further from trees (95 cm from trees) (Table 4).Soil moisture close to trees was most often (August 2012, June 2013, August 2013) greater in the treatment with the removal of both above-and belowground vegetation (BC) as compared to the other treatments, the latter of which did not differ from one another.However, in July 2013, when soil moisture was generally higher, the opposite was true (i.e., moisture was lowest in the above&below-close treatment; Table 3).In August 2012, soil moisture 95 cm from trees was higher with removal of both above-and belowground vegetation than with removal of D r a f t 16 just aboveground vegetation (i.e., above&below-far versus above-far or above-total) but did not differ from the control treatment (Table 4).By July 2013, soil moisture far from trees was lower with removal of both above-and belowground vegetation (BF treatment) compared to all other treatments (Table 4).
Mechanical vegetation (i.e., aboveground) control resulted in significantly higher availability of several nutrients, notably nitrogen, potassium, and phosphorous, both near and far from trees (Tables 3   and 4; recalling we did not assess nutrient availability in treatments removing vegetation both aboveand belowground).Aboveground vegetation removal from 0-140cm (above-total) resulted in the greatest nutrient increases relative to the control treatment (NR) (Tables 3 and 4).

Discussion
Hybrid poplar productivity was improved by the suppression of competing vegetation, thereby highlighting the importance of weed control for improving tree growth in plantations containing abundant herbaceous cover.Our results also revealed marked differences in the effectiveness of vegetation removal methods targeting near-stem (0-50 cm) versus far-stem (50-140 cm), and above-(i.e., partial) versus belowground (i.e., complete: above-+ belowground) control.
First year tree growth was improved through selective in-row vegetation control close to trees, regardless of whether vegetation removal was achieved aboveground (i.e., mechanically) or above-and belowground (i.e., chemically).Notably, no further improvements in tree growth were observed during the first growing season by adding control of vegetation farther from the tree stem.This finding is important given the larger treatment area for the removal farther from trees (5.37 m 2 ) compared to near trees (0.79 m 2 ), and reinforces the importance of relatively intense near-bole competition in regulating the growth of young trees.Similarly, in the second year, control of vegetation close to the stem generally resulted in greater diameter growth than treatments that controlled vegetation farther from the stem.Despite the increase in tree growth associated with control of both above-and D r a f t 17 belowground vegetation, it is worth noting that this benefit was at least partly offset by reduced tree survival in trees that received the herbicide application close (0-50 cm) to the stem.This was particularly evident in Walker poplar, for which survival by the fourth growing season was only 17% for trees in the above&below-close treatment.
The importance of initially controlling vegetation near the tree stem (above-or belowground), as shown in 2011 and 2012, likely reflects the high resource requirements of hybrid poplar during the early establishment period and the strong competitive effects of neighboring herbaceous vegetation (Coll et al. 2007, Kabba et al. 2007, Henkel-Johnson et al. 2016).The fact that both mechanical (aboveground) and chemical (above-and belowground) vegetation control near the tree stem had a similar positive effect suggests that at least part of the benefit to young trees was through an increase in light availability.All treatments that removed vegetation near the tree stem (above-close, above-total, above&below-close) led to increased transmission of light to trees (by ~25%).However, treatments removing both above-and belowground vegetation resulted in the greatest light transmission and proved most effective for longer-term in-row and between-row control of understory vegetation; mechanical control, on the other hand, demonstrated only short-lived effects on competing vegetation, despite the high frequency of treatment applications.These findings concur with previous studies highlighting the effectiveness of repeated herbicide application, as compared to mechanical treatments, for suppressing competing herbaceous vegetation (Siipilehto 2001, Wagner et al. 2005, Balandier et al. 2006, Coll et al. 2007, Morhart et al. 2013).
Vegetation removal close to trees is of particular importance in the presence of tall-growing forbs that have a strong shading effect due to their large leaf area (Balandier et al. 2006, Grenke et al. 2016).Herbaceous vegetation in our study was strongly dominated by annuals, and more importantly, perennial forbs, including the tall growing species Medicago sativa (volunteer alfalfa), Artemisia biennis (biennial wormwood) and Cirsium arvense (Canada thistle, a noxious weed).When left uncontrolled, D r a f t 18 these species were taller than young poplar trees during the first year after planting.Balandier et al. (2009) found that Medicago sativa and Taraxacum officinale (dandelion) can decrease light transmission to a level that causes mortality of small trees.Removal of both above-and belowground vegetation (through repeated herbicide applications) also resulted in the lowest abundance of perennial grasses, and this would benefit tree growth, as even a low cover of perennial grasses is known to reduce hybrid poplar survival and growth (Kabba et al. 2007, Grenke et al. 2016, Henkel-Johnson et al. 2016).
Interestingly, our mechanical treatment resulted in relatively larger reductions in perennial than annual forb cover, and likely reflects the fact that repeated mowing may limit perennial vegetation by removing above-ground meristems, with limited opportunities for regrowth.In contrast, annual forbs appeared to be hyper-abundant in the soil seed bank and were continually emerging throughout the growing season, a rapid process that would have been favored by greater soil temperatures and light at the ground surface following periodic mowing.
There was also evidence that the benefits of vegetation control included belowground resources.Removal of both above-and belowground vegetation resulted in greater soil moisture than the removal of only aboveground vegetation, the latter of which would have left live roots to continue water and nutrient uptake in the treated area.Vegetation control also resulted in increased soil temperature, which has been associated with increased mineralization, tree root activity, and tree growth (Hansen et al. 1986, Nyborg and Malhi 1989, Pinno and Bélanger 2009).We found that mechanical suppression of aboveground vegetation increased nutrient supply, as compared to the nontreated control.Along with increased mineralization, these differences in soil nutrient availability could partly be attributed to reduced nutrient uptake by herbaceous vegetation.Removal of neighboring vegetation throughout the growing season could also have led to root dieback of weedy understory vegetation (Bicksler et al. 2012), thereby leading to greater nutrient availability following root decay.

D r a f t 19
Our findings suggest that, unlike the business-as-usual plantation management approach of repeatedly spraying and cultivating between rows (leaving ~30-50 cm untreated near the stem; Henkel-Johnson et al. 2016), vegetation control farther from the tree did little to enhance tree growth during the first two years after planting.This agrees with the findings by Davies (1988) and Thomas et al. (2001).Greater benefits at that time would be afforded by concentrating weed control on the suppression of vegetation immediately surrounding the stems of newly planted poplar trees, either through more intensive initial weed control before planting (Goehing et al. 2017) or by within-row control after planting, being careful to ensure there is no transfer of herbicide to the main stem.
In contrast, results from the third and fourth year of monitoring showed that removal of aboveand belowground vegetation close to the tree stem did not result in the greatest tree performance; instead, tree growth was maximized when aboveground vegetation was controlled out to a distance of 140 cm from the stem.This finding suggests the competitive effects of neighboring vegetation may shift from near the tree stem to further away (i.e., beyond 50 cm) two years after planting, and thus between-row vegetation control would become more important at that time.These results likely reflect tree root extension farther from the bole over time (Friend et al. 1991); essentially, as the trees grow they become more susceptible to competition for resources at distances farther from the tree.However, if vegetation control further from the tree is not instigated early after planting, implementing between-row cultivation two years after planting could damage tree roots that have grown into that area.Thus, above-and belowground control of competing weeds by means of a herbicide application might be a better alternative because it leaves lateral roots undisturbed, though the latter would only be true if herbicide does not affect trees, including through root uptake via bioactive residue in soil.
Reduced tree growth in the treatment involving near-stem vegetation control with herbicide might reflect unintended herbicide damage, either through accidental application to trees themselves, or through root uptake following the translocation of glyphosate into the soil and subsequent exudation D r a f t 20 into the rhizosphere where it can be taken up by trees (Neumann et al. 2006, Tesfamariam et al. 2009).
Reductions in growth and survival of poplar trees due to herbicide damage have been shown in other studies (i.e., Broeckx et al. 2012), highlighting the challenge of using herbicide applications within plantations, particularly when applied close to trees and during the active growing season (e.g., after leaf-out).Moreover, herbicide applications in this study were done manually, representing a best-case scenario for avoiding incidental contact with trees; in contrast, industrial applications with commercial equipment are more likely to result in non-target chemical application to trees.
While our findings do not challenge current operational practices that strive for between-row weed control, they do refine our knowledge of the silvicultural practices likely to maximize tree growth via the need to minimize near bole competition immediately following planting.The latter is challenging given that conventional plantation practices are not capable of eliminating neighboring vegetation near the tree stem (at any time) because of the risk of damaging tree branches and roots by cultivation equipment and/or herbicide damage.The hand-held weed trimmer we used was better able to control near-stem vegetation than would conventional mowing treatments with operational sized equipment (e.g., tow-behind mowers) (e.g., Coll et al. 2007).Results from our study highlight the need to test innovative complementary weed control practices, with a special emphasis on developing effective near-stem weed control in the early establishment and growth phase, and/or improved weed control in the year prior to plantation establishment (Goehing et al. 2017).
Interestingly, our results also showed that tree growth was impacted by initial tree diameter and height only in the first two years after planting, but not thereafter, indicating growth rapidly became a function of the vegetation control treatments tested rather than initial tree size.These results suggest that longer-term growth responses within commercial plantations can be explained by ongoing silvicultural practices, including weed control.

D r a f t 21
Our results also demonstrated that, despite the close relatedness of the two study clones examined (i.e., Okanese is the progeny of Walker), they differed both in growth and their responses to the vegetation control treatments.Okanese poplar had greater overall productivity and survival, and showed relatively greater increases in tree growth when released from competition, indicating this clone was highly responsive to the vegetation control treatments tested.In contrast, Walker poplar had lower productivity and poorer survival, and did not respond as well to the vegetation control treatments evaluated.
Overall, our results highlight the importance of controlling aboveground vegetation close to trees in the first two years after the establishment of hybrid poplar plantations, and this could be attributed to reduced competition for light.In subsequent years, removal of both above-and belowground vegetation further away (i.e., between-rows) became more important and was associated with reduced competition for moisture and nutrients.Okanese poplar consistently outperformed Walker poplar across all treatments tested, emphasizing its greater potential for deployment in shortrotation-intensive-culture plantations in the Canadian Prairies.Future work should address the need to develop practical operational tools and practices to cost-effectively control vegetation near and far from tree stems without negatively impacting the tree itself.
Results (F-value and significance (P)) from split-split-plot mixed-models examining the effects of site (whole-plot), clone (split-plot), vegetation control treatment (split-split-plot) and their interactions on hybrid poplar basal (i.e., root collar) diameter (DI) and height increment (HI) for the first (2011), second (2012), third (2013), and fourth (2014) growing seasons after planting.Initial tree diameter and height (June 2011) were included as covariates when significant, but otherwise were removed from the final model (denoted by n/a).Significant effects are bolded (p< 0.05).1. Results (F-value and significance (P)) from split-split-plot mixed-models examining the effect of site (whole-plot), clone (split-plot), vegetation control treatment (split-split-plot), and their interactions on: volumetric soil moisture, soil temperature and nutrient supply rates (averaged for the measurements 25 cm and 95 cm from the tree), and relative PAR transmittance; and vegetation cover (0-100 cm from the tree).Measurements were made in the second (2012) and third (2013) growing seasons after planting.Significant effects are bolded (p< 0.05).
(See also Tables 3 and 4).2013) growing season after tree planting as measured 25 cm from the tree, or 0-50 cm for vegetation cover.Values represent mean (SD) of 18 blocks per treatment for soil moisture and temperature, and twelve blocks for soil nutrient supply rates.Different lowercase letters in a row indicate significant differences among treatments (at Bonferroni adjusted α adj = 0.05/6= 0.008 for soil moisture, soil temperature, and vegetation, and α adj = 0.05/3= 0.016 for nutrient supply rates).Also given is the significance (p-value) of the vegetation control treatment (significant effects bolded).Each bar is broken down into the cover by each plant functional group.For more detail on analyses by functional group near (0-50 cm) and far (50-100 cm) from the tree, see Tables 3 and 4.

Fig. 1 .
Fig. 1.Growth of Okanese and Walker poplar trees.Given is the mean (s.e.) diameter increment (mm) (panels A to D) and height increment (cm) (panels E to H) during the first (2011, panels A and E), second (2012, panels B and F), third (2013, panels C and G), and fourth (2014, panels D and H) growing seasons after planting for each of the six vegetation control treatments.Different lowercase letters indicate differences among treatments in a given year by clone combination (at Bonferroni adjusted α adj = 0.05/15= 0.003).For the Walker clone there were no significant differences in diameter increment in 2012, 2013, and 2014, or in height increment in 2012 and 2014.For the Okanese clone, height increment did not differ among treatments in 2011.Treatment abbreviations (see also Fig. 1): NR (control), AF (above-far), AC (above-close), AT (above-total), BF (above&below-far), and BC (above&below-close).

Table 3 .
Environmental variables in the second (2012) and third (

Table 4 .
Environmental variables in the second (2012) and third (2013) growing season after tree planting as measured 95 cm from the tree, or 50-100 cm for vegetation cover.Values represent mean (SD) of 18 blocks per treatment for soil moisture and temperature, and twelve blocks for soil nutrient supply rates.Different lowercase letters in a row indicate significant differences among treatments (at