Jukuri, open repository of the Natural Resources Institute Finland (Luke) All material supplied via Jukuri is protected by copyright and other intellectual property rights. Duplication or sale, in electronic or print form, of any part of the repository collections is prohibited. Making electronic or print copies of the material is permitted only for your own personal use or for educational purposes. For other purposes, this article may be used in accordance with the publisher’s terms. There may be differences between this version and the publisher’s version. You are advised to cite the publisher’s version. This is an electronic reprint of the original article. This reprint may differ from the original in pagination and typographic detail. Author(s): Jiri Pyörälä, Riikka Piispanen, Sauli Valkonen and Sven-Olof Lundqvist Title: Tracheid dimensions of Norway spruce in uneven-aged stands Year: 2021 Version: Final draft Copyright: The Author(s) 2021 Rights: This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights- holder(s) Rights url: https://rightsstatements.org/page/InC/1.0/?language=en Please cite the original version: Mr. Jiri Pyörälä, Dr. Riikka Piispanen, Dr. Sauli Valkonen, and Mr. Sven-Olof Lundqvist. Tracheid dimensions of Norway spruce in uneven-aged stands. Canadian Journal of Forest Research. Just-IN https://doi.org/10.1139/cjfr-2021-0140 11 Tracheid dimensions of Norway spruce in uneven-aged stands 2 3 List of authors: 4 Jiri Pyörälä1), email: jiri.pyorala@helsinki.fi and 5 Riikka Piispanen2), email: ext.riikka.piispanen@luke.fi 6 Sauli Valkonen3), email: sauli.valkonen@luke.fi 7 Sven-Olof Lundqvist4) email: svenolof.lundqvist@indic.se 8 9 For 1) Affiliation: University of Helsinki, Department of Forest Sciences 10 Address: Latokartanonkaari 7, P.O. Box 27, 00014 University of Helsinki 11 For 2) and 3) Affiliation: Natural Resources Institute Finland, Luke 12 Address: Latokartanonkaari 9, P.O. Box 2, FI-00791 Helsinki, Finland 13 For 4) Affiliation: IIC 14 Address: Rosenlundsgatan 48B, 11863 Stockholm, Sweden 15 Corresponding author: Name: Riikka Piispanen 16 Affiliation: Natural Resources Institute Finland, Luke 17 Address: Latokartanonkaari 9, P.O. Box 2, FI-00791 Helsinki, Finland 18 Tel. +358 029 532 5473 19 e-mail: ext.riikka.piispanen@luke.fi Page 1 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 220 Abstract 21 Tracheid length and width patterns from pith to bark at a height of 0.6 m in uneven-aged 22 Norway spruce (Picea abies L. (H.) Karst) trees were addressed. The identification of the 23 main factors and a comparison with even-aged stands were also pursued. 96 trees were 24 sampled from experimental stands in Southern Finland. The material encompassed the 25 variation in tracheid properties from early years to silvicultural maturity, i.e. from corewood 26 to outerwood up to a cambial age of 111 years. Data from 39 Norway spruce trees from even- 27 aged stands we utilized for comparison. Models fitted to the data indicated that annual ring 28 widths did not influence mean tracheid dimensions but the latewood proportion showed a 29 significant influence on tracheid dimensions. Tracheids in uneven-aged stands were slightly 30 wider and longer at the base of the stem with a similar tree diameter, cambial age, and annual 31 ring number. 32 Keywords 33 Wood properties, Norway spruce, tracheid length, tracheid width, uneven-aged stands 34 35 Page 2 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 336 1. Introduction 37 The management of even-aged stands (EAS) has been the overwhelmingly 38 predominant mode of silviculture in the Nordic countries of Finland, Norway, and 39 Sweden since the 1950s, and contemporary forests are characterized by uniform 40 stands with the predominance of a single species with homogenous spacing, stem 41 diameter, height, and canopy characteristics. Clear-cutting is the main method for 42 forest renewal in EAS used throughout the boreal forest zone. Homogeneous stands 43 and large-scale clear-cuts have been increasingly associated with reducing 44 biodiversity and the loss of various ecosystem functions and services (Gauthier et al. 45 2015; Boucher et al. 2017; García-Tejero et al. 2018). 46 In contrast with EAS, ecosystem-based forest management practices such as the 47 management of uneven-aged stands (UAS) are associated with a greater degree of 48 structural heterogeneity which may increase the biodiversity in managed forests 49 (Assmuth and Tahvonen 2018; Nolet et al. 2018). Single-tree and group selection is 50 currently considered the primary UAS method in the management of Norway spruce 51 (Picea abies L. (H.) Karst) dominated stands in the Nordic countries. Uneven-aged 52 stands are made up of trees of multiple ages and sizes, mixed at small spatial scales, 53 resulting in complex competitive interactions between the trees. Selection harvesting 54 is applied, targeting mainly large tree individuals that are considered economically 55 mature. Trees are removed individually or in small groups or patches. Trees typically 56 experience consecutive suppression and release phases, especially at early ages 57 (Schütz 2001). 58 The management of uneven-aged stands is expected to gradually achieve a minor but 59 established status in practical forestry in Nordic boreal forests. There is a great 60 demand for research results on the effects of uneven-aged management on tree growth Page 3 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 461 and wood formation in comparison to those in EAS. Such information is needed for 62 the development of practical guidelines and management services for uneven-aged 63 methods throughout the boreal forest zone (Puettmann et al. 2015), with respect to the 64 expected quantity and quality of wood products produced. 65 The primary wood products in uneven-aged management are sawlogs, and the wood 66 properties and quality of logs and sawn timber from UAS has been previously studied 67 (Seeling 2001; Macdonald et al. 2010; Piispanen et al. 2014; Pretzsch and Rais 2016; 68 Pamerleau-Couture et al. 2019; Piispanen et al. 2020). However, a considerable 69 portion of the harvested volume is used for pulp from culled logs that are unsuitable 70 for producing lumber (an average of 33% was reported by Laamanen (2014)), and 71 from chips produced from the outside of the logs during lumber manufacturing (slabs, 72 chips, and sawdust make up approximately 33% of the sawlog volume according to 73 Natural Resources Institute Finland, Statistics database, https://stat.luke.fi/en/, 74 statistics 2018). Tracheid dimensions are important pulp quality indicators that can be 75 used to predict final pulp properties, for instance, the stiffness and coarseness of pulp 76 fibers and density of sheets produced. 77 The development of tracheid dimensions in xylem exhibits well-established 78 relationships with the cambial aging and radial growth of the stem (Lindström 1997; 79 Fabris 2000; Sirviö and Kärenlampi 2001, Savva et al. 2010). The dependency 80 reflects the processes of wood formation, adapting to changes in water conductivity 81 and mechanical load, regulated through hormonal responses that are also affected by 82 the environment, e.g. silviculture, climate, and canopy position (Schrader et al. 2003; 83 Rathgeber et al. 2011; Sorce et al. 2013). Wood formation and patterns of tracheid 84 size development from pith to bark are in large part genetically encoded (Zobel and 85 Jett 2012), but forest management also influences tree growth and tracheid formation Page 4 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 586 via its effect on the growth conditions. Due to heterogeneous canopy structures and 87 the repeated removal of neighboring trees in UAS, multiple changes in the ranking of 88 the trees’ canopy positions occur throughout the life-cycle of a tree, in contrast with 89 EAS where trees mainly maintain similar canopy positions over the entire rotation. In 90 trees from UAS, reflections of shifting canopy positions can be found in the wood 91 properties. For example, the corewood had narrower annual rings and larger 92 proportions of latewood, while the outermost annual rings were wider and had less 93 latewood (Piispanen et al. 2014). These are likely consequences of slow growth in 94 young trees, suppressed by the older trees, and faster growth in old trees in dominant 95 positions among younger trees. To date, very few studies have investigated tracheid 96 properties in UAS-grown trees (Pamerleau-Couture et al. 2019), and to our 97 knowledge, the effect from early suppression during the corewood production phase 98 on the development of tracheid dimensions from pith to bark has not been studied 99 before. 100 101 The objectives of this study were to 102 1) characterize the patterns of tracheid width and tracheid length along the stem 103 radius (pith-to-bark) in Norway spruce trees in uneven-aged stands; 104 2) identify the effects of varying growth rates (in terms of ring width, and latewood 105 percentage) of tracheid dimensions along the stem radius; 106 3) compare the tracheid dimensions between even-aged and uneven-aged 107 management. 108 109 2. Materials and Methods 110 2.1. UAS data source Page 5 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 6111 Sample trees were harvested from permanent UAS experiments belonging to a set of 112 25 experimental stands at three geographic locations in southern Finland (60.6–62.6 113 N, 25.1–27.1 E): Lapinjärvi (LAP), Vesijako (VES) and Suonenjoki (SUO). Five 114 stands (LAP01 and 13; VES; SUO03 and 04) were selected for studies of wood 115 properties (Piispanen et al. 2014, 2020 and this study). The selection criteria were: i) a 116 balanced coverage of geographic area and site conditions; and ii) stand properties 117 resembling those associated with single-tree selection (i.e. balanced spatial and size 118 distribution of vigorous trees). The stands were located on mineral soil and classified 119 as the submesic Myrtillus site type, except stand LAP01, which was on a mesic 120 Oxalis-Myrtillus site type (Cajander 1949). All were dominated by Norway spruce but 121 also contained admixtures of Scots pine (Pinus sylvestris L.) and various broadleaf 122 species, mainly silver birch (Betula pendula Roth), downy birch (Betula pubescens 123 Ehrh.) and aspen (Populus tremula L.) with a 10–50% proportion of volume. 124 All the stands could at that time be characterized as truly multi-aged (with trees aged 125 up to 170 years present) and full-storied, as defined by Ahlström and Lundqvist 126 (2015). Single-tree selection targeting the maintenance or enhancement of the existing 127 complex stand structures was adopted in the 1980s, and a single-tree selection cutting 128 was carried out in all of the stands in 1985–1988. After the establishment of the 129 experiments in 1991–1996, the selection cutting was repeated in 1996 and 2011 in 130 LAP01 and LAP13, 2007 at VES, and 2008 at SUO03 and SUO04. All trees with 131 defects or damage were removed first, then healthy trees mainly from the larger 132 diameter classes (>25 cm) were removed until a target basal area was achieved. 133 134 Each stand comprised an area of 1–2 ha. One experimental plot was placed in the 135 central area of each stand. In LAP01 and LAP13, each stand had only one plot (40 m Page 6 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 7136 × 40 m) with a single basal area level within the plot (14.2 m2 ha-1 for LAP01 and 137 18.7 for LAP 13, respectively). The stand outside the plot was treated similarly to that 138 inside the plot. In the three stands, VES, SUO03, and SUO04, the plots were larger 139 (80 m × 100 m) and were divided into eight subplots of different basal areas introduce 140 density variation within the stand. A series of four basal area levels was therefore 141 established in two replications on each of these three plots. In VES, the levels were 8, 142 12, 16, and 20 m2 ha-1, while in SUO03 and SUO04, they were 10, 15, 20, and 25 143 m2ha-1. The subplot sizes were 40 m × 20 m or 40 m × 30 m (edge plots). The stand 144 adjoining the subplot on the outside was harvested similarly to the subplot. The 145 sampling for this study was evenly distributed between subplots to ensure a wide 146 representation of stand densities. More details on silvicultural and experimental issues 147 can be found in Saksa and Valkonen (2011), Eerikäinen et al. (2014), and Hynynen et 148 al. (2019). 149 150 A total of 96 Norway spruce trees was sampled (Piispanen et al. (2014),), 32 trees 151 from each experimental stand. The sampling was evenly stratified across subplots in 152 stands where subplots existed (VES, SUO3, SUO4), and the tree diameter classes (or 153 size classes) to ensure even representation. Four trees (eight trees at VES) were taken 154 from each size class (0–9.99 cm, 10–19.99 cm, 20–29.99 cm and > 30 cm stem 155 diameter at breast height). In VES, no trees were available in the size class of 10– 156 19.99 cm on subplots 7 and 8. Consequently, more trees were sampled from subplot 1. 157 2.2. EAS data source 158 A combination of existing Finnish and Swedish datasets was used to constitute 159 comparable EAS data. Five stands from southern Sweden were selected, which 160 represented different levels of soil fertility on mineral soil in normal commercial Page 7 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 8161 forests (57.1–57.2N, 14.8–15.3E; plots SWE 1–5) (Piispanen et al. 2014,). Two of the 162 stands were approaching maturity, and three stands had undergone their first 163 commercial thinning. At least 10 years had elapsed since the last thinning, and the 164 stand basal areas (22–30 m2ha-1 with a dominant height of 15–17 m) were consistent 165 with the guidelines for best practices in Finnish Forestry (Rantala 2011). The stands 166 had been planted, except the oldest one (SWE 5), for which the regeneration method 167 was unknown. A circular plot was subjectively placed at a representative point in each 168 stand. The plot size was adjusted to include at least 25 healthy trees (radius 6–12 m). 169 Tree species and stem diameter at breast height were recorded for each tree. On each 170 plot, the cumulative basal area distribution of the trees was divided into three classes 171 of equal size, and one sample tree was randomly chosen from each class. The sample 172 trees had to be healthy, i.e. no visible damage was allowed. 27 trees were sampled. 173 174 In Finland, the EAS materials were collected from two thinning experiments in 175 Heinola and Parkano (FIN 6 and FIN 8), in southern Finland (61.2–62.2N, 22.9– 176 26.0E). The stands represented mesic Oxalis-Myrtillus and submesic Myrtillus site 177 types (Cajander 1949). Experiments FIN 6 and FIN 8 were established in 1970 and 178 1966, respectively. Both stands had been planted and were at the final cut stage at the 179 time of this study. In both experiments, a plot representing the stand density 180 recommended in best practices was selected (Valkonen 2011). The cumulative basal 181 area distributions of the trees were divided into three classes of equal size, and two 182 sample trees were randomly sampled from each class. 12 trees were sampled. Detailed 183 experimental design and stand descriptions are given in Piispanen et al. (2014). 184 185 2.3. Measurement of tracheid width Page 8 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 9186 Disks were sawn at a height of 0.6 m and 1 m from each UAS and EAS tree 187 respectively. Radial sample bars with cross-sections approximately 1 cm x 1 cm were 188 sawn from the pith to the bark along the northern radius. The bars were air-dried, after 189 which a radial sample strip (2 mm tangentially and 7 mm longitudinally) was sawn 190 from each bar. They were extracted with acetone in a Soxhlet Extractor at 56.2 oC for 191 6 h, and their upper surfaces were finely polished. They were stored in the 192 conditioned measurement laboratory for even moisture content before characterization 193 of the radial variations in several tracheid and wood properties from pith to bark with 194 the SilviScan instrument (Innventia Ab, Stockholm, Sweden) (Evans et al. 1995). 195 SilviScan integrates three measurement principles: image analysis of tracheid cross- 196 sections; X-ray absorption to measure wood density; and X-ray diffraction to measure 197 the microfibril angle of wood strips (Evans 1994; Evans and Elic 2001). In this study, 198 image analysis and X-ray absorption were used for measuring tracheid width, widths 199 in the radial and tangential directions, in radial intervals of 25 µm (Piispanen et al. 200 2014). 201 The locations of all annual rings and the interfaces between their parts of earlywood, 202 transition wood, and latewood were determined from the radial variations in wood 203 density (Lundqvist et al. 2018). The ring boundaries were determined using an 204 algorithm that detected the steep density decrease that occurs between the latewood 205 and the earlywood. As many annual rings in UAS trees were exceptionally narrow, 206 cross-checking was conducted to assure the locations of annual ring boundaries 207 optically on samples taken perpendicular to those for the SilviScan measurements 208 (Piispanen et al. 2014). The same samples were remeasured using a computer-aided 209 system consisting of an Olympus SZ51 stereo microscope (Olympus corporation, Page 9 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 10 210 Tokyo, Japan) connected to a Heidenhein LS 303C transducer (Encoders UK Ltd. 211 Birmingham, UK) with an accuracy of 0.001 mm (Piispanen et al. 2014). 212 After the setting of ring boundaries and the cross-checking, a “20–80” density 213 threshold definition was applied for separating three compartments within each ring. 214 The span from minimum to maximum wood density was determined based on the 215 SilviScan measurements: The part from 0 to 20% of the span was classified as 216 earlywood, the part from 80 to 100% as latewood, and the part between as transition 217 wood (Lundqvist et al. 2018). 218 219 From the locations, ring distance from pith (stem radius, R), ring number from the 220 pith (cambial age, CA), ring widths (RW), latewood and transitionwood widths and 221 proportions (LWW, TWW, LWP and TWP, respectively) were calculated. Arithmetic 222 means of radial tracheid widths was calculated for each ring (Width), as well as for its 223 earlywood (WidthEW), latewood (WidthLW) and transitionwood (WidthTW). A total 224 of 6004 and 1191 annual rings in the UAS and EAS trees were measured, 225 respectively. 226 The first five rings were excluded as the innermost rings are too curved to allow 227 precise data acquisition for each ring and its parts with the x-ray beam passing 228 through the 2 mm thick sample strip. 229 2.4. Measurement of tracheid length 230 Disks were sawn just below a height of 0.6 m and 1.0 m from each UAS and EAS tree 231 respectively. In UAS, only trees from sites LAP and VES were included. A radial 232 strip (0.5 cm tangentially and 1.5 cm longitudinally) was sawn from the pith to the 233 bark along the north radius from each disk. For UAS trees, one-centimeter-long radial 234 samples were cut from the 0.5-cm-wide strip from pith to cambium and cut into Page 10 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 11 235 match-size sticks for maceration. For UAS trees, the R and CA of the samples refer to 236 the values of the rings of the oldest cambial age of the 1-cm section of the radial 237 samples, and RW, LWW, and TWW were defined as sample-specific averages. For 238 EAS trees, earlywood samples were taken from selected single annual rings and cut 239 into match-size sticks for maceration. After analysis, the means of the tracheid lengths 240 in the EW samples were calculated. In EAS trees, R and CA referred to the exact 241 rings. A total of 754 and 30 samples with 50 tracheids each was gained for UAS and 242 EAS trees respectively. 243 Prior to the analyses of tracheid length, the samples were macerated in glacial acetic 244 acid/30% hydrogen peroxide (1:1, v/v) overnight at 60 °C. The resulting suspension 245 of the liberated washed fibers of each sample was applied to four microscope slides. 246 Fifty unbroken tracheids per sample were measured with an Olympus BH-2 247 microscope (Olympus Optical, Tokyo, Japan) and a CCDC Camera (COHU MOD 248 4912-5000/0000, Cohu, San Diego, CA) in conjunction with the Image-Pro Plus 3.0 249 program for Windows (Media Cybernetics, Silver Spring, MD). One pixel 250 corresponded to 1.08 µm. The mean of the tracheid lengths measured was used to 251 describe the tracheid length (L). 252 253 2.5. Statistical analysis 254 We analyzed the variation of tracheid Width and Length in UAS and EAS, using 255 modeling as the primary statistical approach. Linear mixed modeling was applied to 256 account for the random variances arising from the hierarchical structure of the data 257 (Dutilleul et al. 1998; Downes et al. 2002). Variance components (VC) for plot, tree, 258 and growth ring levels were incorporated in the mixed models to allow individual Page 11 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 12 259 observations to vary around the population, plot, and tree means respectively. First- 260 order autoregression (AR(1)) was applied to adjacent rings, as they are often 261 correlated (Dutilleul et al. 1998; Downes et al. 2002). 262 The data from the first five annual rings were excluded from both modeling datasets, 263 according to the limitations in measuring the narrow rings near the pith (see above). 264 Upon screening the linear mixed model candidates, we found that one tree had 265 exceptionally small tracheids, probably due to a very high compression wood content, 266 and that tree was excluded from the modeling data. The eventual number of samples 267 used for modeling Width and Length was then 5598 and 709 respectively. 268 Because our response variables were mean values of entire growth rings (or several 269 rings in the case of Length of UAS trees), they were modeled with respect to ring- 270 level attributes: R, CA, RW, and the widths and proportions of LW and TW (LWW, 271 TWW, LWP, and TWP). 272 The growth trends of cell dimensions in the radial direction are well established - 273 several studies have used e.g. logarithmic or exponential functions to describe the 274 pith-to-bark developments (Lundqvist et al. 2005; Franceschini et al. 2012; Piispanen 275 et al. 2014). In this study, we used logarithmic transformations of the response 276 variables, and R and CA were the first to be tested in models for both dimensions. 277 Subsequently, RW and its various transformations were tested in the models, with the 278 assumption that they could explain additional variance along the radial trend caused 279 by differences in growth conditions, tree position, and other factors affecting RW. In 280 addition, LWP (and TWP, to a lesser extent) is known to be an important explanatory 281 variable for the ring-specific means of tracheid dimensions, due to LW tracheid 282 properties that differ from those in EW (Lindström 1997). It has also been argued that Page 12 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 13 283 LWP (often negatively correlated with RW) could actually express most of the 284 variance often attributed to RW (Downes et al. 2002). We therefore also tested LWP 285 and TWP in our models. 286 Relative stem size (dbh/dbhdom), where dbh is the individual tree stem diameter at 287 breast height and dbhDom is the average stem diameter of the 100 thickest trees ha-1) 288 was tested as a fixed variable explaining tree-specific variations of tracheid 289 dimensions. Site was tested as a fixed variable to account for the differences between 290 the studied sites that were not captured by the selected explanatory variables. 291 Following the screening of various combinations of the explanatory variables and 292 their transformations and mutual correlations (Tables A1 and A2 in Supplementary 293 material), final models were selected based on visual inspections. Model fit and 294 unbiasedness were ensured by the visual examination of model residuals with respect 295 to the observed values, as well as the explanatory variables, and the statistical 296 significance of the parameter estimates. The predicted response variable values ln (𝑦) 297 were retransformed to the original scale (µm and mm for Width and Length 298 respectively) using the following equation: 299 (1),𝑦 = 𝑒ln (𝑦) + (𝛿𝑘𝑗 + 𝛼𝑗 )/2 300 where δkj and αj are the random terms for tree- and plot-level effects, respectively. The 301 correction term was added to the estimates of the intercepts. 302 303 Differences of tracheid dimensions between EAS and UAS were examined by 304 applying the UAS models (2 and 3) to the data for EAS and analyzing the errors. 305 All statistical analyses were carried out using SPSS© version 25 (IBM 2018). Page 13 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 14 306 307 3. Results 308 3.1. Empirical results 309 Both tracheid dimensions increased logarithmically with greater R and CA (Figures 2, 310 3). In UAS, differences in the tracheid dimensions between the size classes of trees 311 were notable with respect to CA, while the differences were less pronounced with 312 respect to R. 313 Tracheid Width showed a larger range of variation in UAS trees of different sizes than 314 in EAS trees, in general and when related to R, CA, RW, and LWP (Figure 2, Table 2). 315 A major difference between the two treatments was observed for the innermost 10 316 annual rings, where the UAS trees had a much steeper increase in the tracheid widths, 317 synchronous to the rapid decrease in LWP (Figure 1). 318 In UAS, tracheid Length was largely independent of RW. However it differed among 319 the size classes (Figure 3, Table 2), and had a negative correlation with LWP. 320 3.2. Tracheid dimension models 321 The finally selected tracheid dimension models were: 322 ln 𝑊𝑖𝑑𝑡ℎ 𝑖𝑗𝑘𝑙 = 𝑏0 + 𝑏1 ∗ 𝑆𝑖𝑡𝑒𝑖 + 𝑏2 ∗ ln 𝑅𝑖𝑗𝑘𝑙 + 𝑏3 ∗ 𝐶𝐴2𝑖𝑗𝑘𝑙 + 𝑏4 ∗ 𝑅𝑊 𝑖𝑗𝑘𝑙 + 𝑏5 ∗ 𝑅𝑊2𝑖𝑗𝑘𝑙 323 (2)+ 𝑏6 ∗ ln 𝐿𝑊𝑃𝑖𝑗𝑘𝑙 + 𝑏7 ∗ 𝑙𝑛𝑇𝑊𝑃𝑖𝑗𝑘𝑙 + 𝜀𝑖𝑗 + 𝜀𝑖𝑗𝑘 + 𝜀𝑦𝑒𝑎𝑟𝑙 + 𝜀𝑖𝑗𝑘𝑙 324 ln 𝐿𝑒𝑛𝑔𝑡ℎ𝑖𝑗𝑘𝑚 = 𝑏0 + 𝑏1 ∗ 𝑆𝑖𝑡𝑒𝑖 + 𝑏2 ∗ ln 𝑅𝑖𝑗𝑘𝑚 + 𝑏3 ∗ 𝑙𝑛𝐶𝐴𝑖𝑗𝑘𝑚 + 𝑏4 ∗ 𝐶𝐴2𝑖𝑗𝑘𝑚 + 𝑏5 ∗ 325 (3)𝑅𝑊 𝑖𝑗𝑘𝑚 + 𝑏6 ∗ ln 𝐿𝑊𝑃𝑖𝑗𝑘𝑚 + 𝑏7 ∗ 𝑙𝑛𝑇𝑊𝑃𝑖𝑗𝑘𝑚 + 𝜀𝑖𝑗 + 𝜀𝑖𝑗𝑘 + 𝜀𝑦𝑒𝑎𝑟𝑚 + 𝜀𝑖𝑗𝑘𝑚 326 Readers are referred to the parameter definitions in Table 1, except: Page 14 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 15 327 i, j, k, l, m = Indicators of hierarchy in the data: annual ring l in tree k 328 on plot j at site i; 329 in model 3 for L, Rijkm and CAijkm represent the edge furthest from the pith of the 1-cm 330 thick sample m in tree k on plot j at site i, and other fixed variables in model 3 are 331 averages in the 1-cm thick sample; 332 εij = Random effect of plot within site 333 εijk = Random effect of tree on plot within site 334 εyearl or εyearm = Random effect of calendar year (when the annual ring was 335 established), also including an autocorrelation parameter ρijkl with the AR(1) structure 336 between successive calendar years 337 εijkl or εijkm = Residual. 338 In total, the models explained 92.4% and 84.2% of the variance of Width and Length 339 in UAS data respectively. We found R and RW the most important predictors of Width 340 (Table 3), while Length was mostly explained by R and CA (Table 4). The combined 341 effects of the independent variables CA, RW, and LWP on Width and Length in trees 342 under UAS management were simulated across relevant ranges and are illustrated in 343 Figure 4. 344 In general, our model (Equation 2) predicted increasing Width for UAS trees with 345 ageing cambium, increasing RW, and decreasing LWP (Table 3). More specifically, 346 the effects of RW and LWP were limited during the juvenile phase (CA < 20) but 347 became increasingly influential as the cambium aged (Figure 4). The most distinct 348 enlargement of Width was predicted with CA> 60 a, RW> 4 mm and LWP<10 (Figure 349 4). Page 15 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 16 350 According to the model (Equation 3), Length increased with CA and showed a 351 negligible decrease with increasing RW and LWP, most pronouncedly at ages between 352 30 and 50 (Table 4), after which the influence of CA fully dominated (Figure 4). 353 354 355 3.3. Comparison between UAS and EAS 356 The application of the models (Equations 2 and 3) to EAS data—i.e. predictions of 357 tracheid dimensions in EAS trees using the parameter estimates fitted to UAS data 358 (Tables 3 and 4)—resulted in notable prediction errors in Width and Length. Smaller 359 values were underestimated and larger values overestimated, and the overestimation 360 increased toward the stem surface (Figure 5). These errors reflected the smaller range 361 of Width values in the EAS data (Figure 2), but also the trend of increasing RW 362 synchronously to decreasing LWP towards stem surface (Figure 1) that was associated 363 with increasing Width in the UAS data but was absent in the EAS data (Figure 5). The 364 similar prediction errors of Length were slightly affected by the differences in the 365 growth rates between EAS and UAS (Figure 5). 366 367 368 4. Discussion 369 Tracheid length and width patterns from pith to bark in uneven-aged Norway spruce 370 (Picea abies L. (H.) Karst) trees were addressed. The identification of main factors 371 and a comparison with even-aged stands were also pursued. Page 16 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 17 372 The wood samples were taken near the stem base (0.6 m) to cover the full range of the 373 annual ring width patterns in uneven-aged stands. However, the study material cannot 374 be considered representative of tracheid properties at other heights along the stem. 375 The difference in sampling heights (1.0 m for EAS vs. 0.6 m for UAS) should not 376 have had a major influence on the results The sample size for tracheid length for EAS 377 (30 samples, 1500 tracheids) is not large enough to yield valuable insight into the 378 patterns beyond general trends. 379 We observed patterns of tracheid dimensions developments that reflected the effects 380 of the initial suppression in the early years, and the gradual release toward the 381 dominant phase. In the UAS material, the latewood proportions in the innermost rings 382 were at relatively high levels (~30%), followed by a steep decrease. Correspondingly 383 rapid increases in the tracheid dimensions in the corewood of the UAS trees were also 384 observed. On the other hand, low latewood percentages and large ring widths were 385 observed in the UAS outerwood, where the tracheids were wider and longer. 386 Both cambial age and stem radius were included in the final models despite their 387 obvious autocorrelation, as their counteraction reflected the effects of shifting canopy 388 positions characteristic of trees grown under the UAS regime. The intra-annual 389 variations of tracheid dimensions due to differing compositions of early-, transition-, 390 and latewood were accounted for in our models by including ring width, latewood, 391 and earlywood proportion in the fixed parts. Their application in the model estimation 392 and application also accounted for the different sampling designs used in the UAS and 393 EAS study materials: tracheid length for UAS trees was measured from 1-cm radial 394 samples entailing entire growth rings including early-, transition-, and latewood 395 proportions, while that of EAS was measured from pure earlywood samples. Page 17 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 18 396 In our models, ring width had more pronounced positive effects onto tracheid widths 397 in the outerwood than in the corewood, and showed only negligible, negative effects 398 on tracheid lengths in addition to the dominant effects of cambial age and stem radius. 399 It is noteworthy that total tree age has been found to be positively correlated with the 400 rate of cell division (Rathgeber et al. 2011; Lundqvist et al. 2018), which is a main 401 determinant of ring width. Subsequently, the observed positive effects of ring width 402 on tracheid width could also be corollary to increasing tree age and stem radius, rather 403 than causal per se. 404 Based on our data and literature, corewood in EAS trees exhibits large ring widths 405 with high earlywood contents, and the proportion of corewood in stems is high 406 (Downes et al. 2002; Sarén et al. 2004; Lundqvist et al. 2005; Lasserre et al. 2009). 407 However, we observed contrasting trends in the UAS trees. The differences may be 408 related to the maturation processes in cambium due to the long period of slow growth 409 resulting from the suppression phase typical in UAS. Most UAS trees in our data had 410 thinner rings and higher latewood proportion in their corewood, as opposed to the 411 wider rings with a higher earlywood proportion in the EAS corewood. At a similar 412 stem radius, the cambium in UAS was thus generally older than in EAS (i.e. had 413 higher cambial age). Tracheids from dominant UAS trees (DBH≥30 cm; wider rings 414 and lower latewood proportion ) reached larger dimensions at smaller stem radius and 415 younger cambial age compared to EAS. The outerwood in dominant UAS trees had 416 wider tracheids than those in EAS. In contrast, tracheids from the outerwood of 417 suppressed UAS trees (DBH<10 cm; thinner rings and higher latewood proportion), 418 remained narrower than their EAS counterparts. 419 Despite the differences, our results concur with similar observations about the age- 420 and size-mediation of tracheid dimension maturation in the major body of the Page 18 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 19 421 previous literature, e.g. Lindström (1997); Mäkinen et al. (2007); Franceschini et al. 422 (2012) and Sirviö and Kärenlampi (2001). The predominant relationships of tracheid 423 dimensions with cambial age and stem radius reflect the structural-functional 424 relationship of tracheid dimensions with turgor pressure and hydraulic resistance in 425 water transport, which change as the tree ages and grows in height and crown size 426 (Mansfield et al. (2007), Kuprevicius et al. (2013), Sorce et al. 2013). Large UAS 427 Norway spruces may maintain long vigorous crowns when in dominant positions 428 (Kumpu et al. 2020), and produce wood with high earlywood content and wide 429 tracheids at the base of the stem to sustain the water conductivity. 430 In a study of the wood properties of black spruce (Picea mariana (Mill.) in even- and 431 uneven-aged boreal stands in Eastern Canada, Pamerleau-Couture et al. (2019) 432 concluded that most measured wood properties were correlated with ring width, and 433 there were major differences between even-aged and uneven-aged stands submitted to 434 partial cutting. Ring width, tracheid length, latewood and maximum wood density 435 were higher in the even-aged trees than in the uneven-aged trees. These findings 436 diverge partly from those of this study and those of Piispanen et al. (2014) on wood 437 density in our experimental stands. However, direct comparisons are not very 438 meaningful due to the large differences between the materials (geographical location, 439 tree species, intensity and type of cuttings, time elapsed after cuttings). 440 In conclusion, the development of tracheid dimensions between UAS and EAS trees 441 follows similar age and size-dependent relationships. However, the extended 442 suppression phase in the early growth of UAS trees tends to increase the proportion of 443 wider and longer tracheids at the base of the stem. Moreover, old UAS trees in 444 dominant positions produce wood with a high earlywood content. The results suggest 445 that extending the growth of mature trees before harvesting increases the recovery of Page 19 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 20 446 wood with properties well-suited for most end uses. In UAS, the conditions for 447 prolonging the lifespan of trees may be superior to those in EAS, due to the increased 448 ring widths (as a proxy for growth rate) in the wood of the dominant UAS trees. Our 449 results mainly apply to the lower parts of trees, which constitute the main product of 450 UAS as valuable sawlogs. The outermost parts are often used in pulping, for which 451 their fiber properties seem very good. 452 Acknowledgements 453 We gratefully acknowledge the technical skills of Pekka Helminen, Tapio Järvinen, 454 Juhani Korhonen, Tapio Nevalainen, and Hilkka Ollikainen (Natural Resources 455 Institute Finland) and Thomas Grahn, Åke Hansson, and Lars Olsson (RISE Innventia 456 AB, Sweden). Södra Cell is acknowledged for wood material from Swedish Norway 457 spruce even-aged sites. The EAS data reused for this study had originally been created 458 in terms of the project “Multi-sectorial database, model system and case studies, 459 supporting innovative use of wood and fibres” (Innovood), which belonged to the 460 Wood Material Science and Engineering Finnish–Swedish Research Program, which 461 had been funded by BUSINESS FINLAND (formerly Tekes) and VINNOVA on the 462 Swedish side. We thank Harri Mäkinen for kindly providing the EAS tracheid length 463 data. 464 Competing interests statement 465 The authors declare there are no competing interests. 466 References Page 20 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 21 467 Ahlström, M.A., and Lundqvist, L. 2015. Stand development during 16–57 years in partially 468 harvested sub-alpine uneven-aged Norway spruce stands reconstructed from 469 increment cores. For. Ecol. Manage. 350, 81-86. doi:10.1016/j.foreco.2015.04.021. 470 Assmuth, A., and Tahvonen, O. 2018. Optimal carbon storage in even-and uneven-aged 471 forestry. For. Policy Econ. 87, 93-100. doi:10.1016/j.forpol.2017.09.004. 472 Boucher, Y., Perrault-Hébert, M., Fournier, R., Drapeau, P., and Auger, I. 2017. Cumulative 473 patterns of logging and fire (1940–2009): consequences on the structure of the eastern 474 Canadian boreal forest. Landscape Ecol. 32 (2), 361-375. doi:10.1007/s10980-016- 475 0448-9. 476 Cajander, A.K. 1949. Forest types and their significance. Acta For. Fenn. 56 (4), 1-69. 477 Downes, G.M., Wimmer, R., and Evans, R. 2002. Understanding wood formation: gains to 478 commercial forestry through tree-ring research. Dendrochronol. 20 (1-2), 37-51. 479 doi:10.1078/1125-7865-00006. 480 Dutilleul, P., Herman, M., and Avella-Shaw, T. 1998. Growth rate effects on correlations 481 among ring width, wood density, and mean tracheid length in Norway spruce (Picea 482 abies). Can. J. For. Res. 28 (1), 56-68. 483 Eerikäinen, K., Valkonen, S., and Saksa, T. 2014. Ingrowth, survival and height growth of 484 small trees in uneven-aged Picea abies stands in southern Finland. For. Ecosys. 1 (1), 485 5. doi:10.1186/2197-5620-1-5. 486 Evans, R. 1994. Rapid measurement of the transverse dimensions of tracheids in radial wood 487 sections from Pinus radiata. Holzforsch. 48 (2), 168-172. 488 doi:10.1515/hfsg.1994.48.2.168. 489 Evans, R., Downes, G., Menz, D., and Stringer, S. 1995. Rapid measurement of variation in 490 tracheid transverse dimensions in a radiata pine tree. Appita J. 48 (2), 134-138. 491 https://search.informit.org/doi/epdf/10.3316/informit.252935342865527. Page 21 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 22 492 Evans, R., and Elic, J. 2001. Rapid prediction of wood stiffness from microfibril angle and 493 density. For. Prod. J. 51 (3), 53-57. 494 Fabris, S. 2000. Influence of cambial ageing, initial spacing, stem taper and growth rate on 495 the wood quality of three coastal conifers. Ph.D. Thesis. University of British 496 Columbia. Canada. doi: 10.14288/1.0089742. 497 Franceschini, T., Lundqvist, S.-O., Bontemps, J.-D., Grahn, T., Olsson, L., Evans, R., and 498 Leban, J-M. 2012. Empirical models for radial and tangential fibre width in tree rings 499 of Norway spruce in north-western Europe. Holzforsch. 66 (2), 219-230. 500 doi:10.1515/HF.2011.150. 501 García-Tejero, S., Spence, J.R., O’Halloran, J., Bourassa, S., and Oxbrough, A. 2018. Natural 502 succession and clearcutting as drivers of environmental heterogeneity and beta 503 diversity in North American boreal forests. PloS ONE, 13 (11), e0206931. | 504 doi:10.1371/journal.pone.0206931. 505 Gauthier, S., Bernier, P., Kuuluvainen, T., Shvidenko, A., and Schepaschenko, D. 2015. 506 Boreal forest health and global change. Sci. 349 (6250), 819-822. 507 doi:10.1126/science.aaa9092. 508 Hynynen, J., Eerikäinen, K., Mäkinen, H., and Valkonen, S. 2019. Growth response to 509 cuttings in Norway spruce stands under even-aged and uneven-aged management. 510 For. Ecol. Manage. 437, 314-323. doi:10.1016/j.foreco.2018.12.032. 511 Kumpu, A., Piispanen, R., Berninger, F., Saarinen, J., and Mäkelä, A. 2020. Biomass and 512 structure of Norway spruce trees grown in uneven-aged stands in southern Finland. 513 Scan. J. For. Res. 35, 252-261. doi:10.1080/02827581.2020.1788138. 514 Kuprevicius, A., Auty, D., Achim, A., and Caspersen, J.P. 2013. Quantifying the influence of 515 live crown ratio on the mechanical properties of clear wood. For. 86 (3), 361-369. 516 doi:10.1093/forestry/cpt006. Page 22 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 23 517 Laamanen, V. 2014. Structure of selection cutting stands, logging conditions, logging costs 518 and logging damages. Master's thesis, University of Helsinki. 85 + 5 p. In Finnish 519 with English abstract. 520 Lasserre, J.-P., Mason, E.G., Watt, M.S., and Moore, J.R. 2009. Influence of initial planting 521 spacing and genotype on microfibril angle, wood density, fibre properties and 522 modulus of elasticity in Pinus radiata D. Don corewood. For. Ecol. Manage. 258 (9), 523 1924-1931. doi:10.1016/j.foreco.2009.07.028. 524 Lindström, H. 1997 Fiber length, tracheid diameter, and latewood percentage in Norway 525 spruce: development from pith outward. Wood Fiber Sci. 29 (1), 21-34. 526 https://wfs.swst.org/index.php/wfs/article/view/2081 527 Lundqvist, S.-O., Grahn, T., Hedenberg, Ö. 2005. Models for fibre dimensions in different 528 softwood species. Simulation and comparison of within and between tree variations 529 for Norway and Sitka spruce, Scots and Loblolly pine. In: 5th workshop on “Wood 530 quality modelling”, IUFRO Working Party 5.01.04, Auckland, New Zealand, Nov 531 22–27 2005. 532 Lundqvist, S.-O., Seifert, S., Grahn, T., Olsson, L., García-Gil, M.R., Karlsson, B., and 533 Seifert, T. 2018. Age and weather effects on between and within ring variations of 534 number, width and coarseness of tracheids and radial growth of young Norway 535 spruce. Eur. J. For. Res. 137 (5), 719-743. doi:10.1007/s10342-018-1136-x. 536 Macdonald, E., Gardiner, B., and Mason, W. 2010. The effects of transformation of even- 537 aged stands to continuous cover forestry on conifer log quality and wood properties in 538 the UK. Forestry, 83 (1), 1-16. doi:10.1093/forestry/cpp023. 539 Mäkinen, H., Jaakkola, T., Piispanen, R., and Saranpää, P. 2007. Predicting wood and 540 tracheid properties of Norway spruce. For. Ecol. Manage. 241 (1), 175-188. 541 doi:10.1016/j.foreco.2007.01.017. Page 23 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 24 542 Mansfield, S.D., Parish, R., Goudie, J.W., Kang, K.-Y., and Ott, P. 2007. The effects of 543 crown ratio on the transition from juvenile to mature wood production in lodgepole 544 pine in western Canada. Can. J. For. Res. 37 (8), 1450-1459. doi:10.1139/X06-299. 545 Natural Resources Institute Finland. 2018. Statistics database, https://stat.luke.fi/en/uusi- 546 etusivu. 547 Nolet, P., Kneeshaw, D., Messier, C., and Béland, M. 2018. Comparing the effects of 548 even‐and uneven‐aged silviculture on ecological diversity and processes: A review. 549 Ecol. Evol. 8 (2), 1217-1226. doi:10.1002/ece3.3737. 550 Pamerleau-Couture, É., Rossi, S., Pothier, D., and Krause, C. 2019. Wood properties of black 551 spruce (Picea mariana (Mill.) BSP) in relation to ring width and tree height in even- 552 and uneven-aged boreal stands. Ann. For. Sci. 76 (2), 43. doi:10.1007/s13595-019- 553 0828-9. 554 Piispanen, R., Heinonen, J., Valkonen, S., Mäkinen, H., Lundqvist, S.-O., and Saranpää, P. 555 2014. Wood density of Norway spruce in uneven-aged stands. Can. J. For. Res. 44 556 (2), 136-144. doi:10.1139/cjfr-2013-0201. 557 Piispanen, R., Heikkinen, J., and Valkonen, S. 2020. Deformations of boards from uneven- 558 aged Norway spruce stands. Eur. J. Wood Wood Prod. 78, 533-544. 559 doi:10.1007/s00107-020-01524-x. 560 Pretzsch, H. and Rais, A. 2016. Wood quality in complex forests versus even-aged 561 monocultures: review and perspectives. Wood Sci. Technol. 50 (4), 845-880. 562 doi:10.1007/s00226-016-0827-z. 563 Puettmann, K.J., Wilson, S.McG., Baker, S.C., Donoso, P.J., Drössler, L., Amente, G., 564 Harvey, B.D., Knoke, T., Lu, Y., Nocentini, S., Putz, F.E., Yoshida, T., and Bauhus, 565 J. 2015. Silvicultural alternatives to conventional even-aged forest management-what 566 limits global adoption? For. Ecosyst. 2 (1), 8. doi:10.1186/s40663-015-0031-x. Page 24 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 25 567 Rantala, S. 2011. Finnish forestry practice and management. Metsäkustannus, Helsinki, 568 Finland. 271 p. 569 Rathgeber, C.B.K., Rossi, S., and Bontemps, J.-D. 2011. Cambial activity related to tree size 570 in a mature silver-fir plantation. Ann. Bot. 108 (3), 429-438. doi:10.1093/aob/mcr168. 571 Saksa, T., and Valkonen, S. 2011. Dynamics of seedling establishment and survival in 572 uneven-aged boreal forests. For. Ecol. Manage. 261 (8), 1409-1414. 573 doi:10.1016/j.foreco.2011.01.026. 574 Sarén, M.-P., Serimaa, R., Andersson, S., Saranpää, P., Keckes, J., and Fratzl, P. 2004. Effect 575 of growth rate on mean microfibril angle and cross-sectional shape of tracheids of 576 Norway spruce. Trees, 18 (3), 354-362. doi:10.1007/s00468-003-0313-8. 577 Savva, Y., Koubaa, A., Tremblay, F., and Bergeron, Y. 2010. Effects of radial growth, tree 578 age, climate, and seed origin on wood density of diverse jack pine populations. Trees, 579 24, 53–65. doi: 10.1007/s00468-009-0378-0. 580 Schrader, J., Baba, K., May, S., Palme, K., Bennett, M., Bhalerao, R., and Sandberg, G. 2003. 581 Polar auxin transport in the wood-forming tissues of hybrid aspen is under 582 simultaneous control of developmental and environmental signals. PNAS, 100 (17), 583 10096-10101. doi:10.1073/pnas.1633693100. 584 Schütz, J.-P. 2001. Opportunities and strategies of transforming regular forests to irregular 585 forests. For. Ecol. Manage. 151 (1-3), 87-94. doi:10.1016/S0378-1127(00)00699-X. 586 Seeling, U. 2001. Transformation of plantation forests – expected wood properties of Norway 587 spruce (Picea abies (L.) Karst.) within the period of stand stabilisation. For. Ecol. 588 Manage. 151(1-3): 195-210. doi:10.1016/S0378-1127(00)00708-8. 589 Sirviö, J., and Kärenlampi, P. 2001. The effects of maturity and growth rate on the properties 590 of spruce wood tracheids. Wood Sci. Technol. 35 (6), 541-554. 591 doi:10.1007/s002260100119. Page 25 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 26 592 Sorce, C., Giovannelli, A., Sebastiani, L., and Anfodillo, T. 2013. Hormonal signals involved 593 in the regulation of cambial activity, xylogenesis and vessel patterning in trees. Plant 594 Cell Rep. 32 (6), 885-898. doi:10.1007/s00299-013-1431-4. 595 Valkonen, S. 2011 Tree species. In Finnish Forestry-Practice and Management. Edited by S. 596 Rantala. Metsäkustannus, Helsinki, Finland. pp. 35-36. 597 Zobel, B.J., and Jett, J.B. 1995. Genetics of wood production. In Springe Series of Wood 598 Science. 1st ed. Springer-Verlag Berlin Heidelberg, Germany. 337 p. 599 Page 26 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 27 600 601 Table 1. List of variables, their abbreviations, explanations, and units. 602 Table 2. Averages, standard deviations and minimum and maximum values for radial 603 tracheid width and tracheid length by tree size classes. 604 Table 3. Parameter coefficient estimates for the linear mixed model of radial tracheid 605 width (Eq. 2), and standard errors (SE) and statistical significances (p). For 606 abbreviations see Table 1. 607 Table 4. Parameter coefficient estimates for the linear mixed model of tracheid length 608 (Eq. 3), and the standard errors (SE) and statistical significances (p). For 609 abbreviations see Table 1. 610 Table A1. Pearson correlation coefficients (r) and significance levels (p) between key 611 variables in the data used for modeling tracheid width. 612 Table A2. Pearson correlation coefficients (r) and significance levels (p) between key 613 variables in the data used for modeling tracheid length. 614 615 Page 27 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . 28 616 Figure captions 617 Figure 1. Observed ring widths and latewood and transition wood percentages in 618 uneven-aged and even-aged stands (UAS and EAS) with respect to cambial age in the 619 four size classes. Lines indicate class means at 15 equal intervals with respect to the 620 x-axis. The data correspond to those used in the measurement of the tracheid widths: 621 for UAS and EAS n = 6004 and n = 1191 respectively. 622 Figure 2. Observed tracheid widths in uneven-aged and even-aged stands (UAS and 623 EAS) with respect to stem radius, cambial age, ring width and latewood percentage in 624 the four size classes. Lines indicate class means at 15 equal intervals with respect to 625 the x-axis. For UAS and EAS, n = 6004 and n = 1191, respectively. 626 Figure 3. Observed tracheid lengths in uneven-aged and even-aged stands (UAS and 627 EAS) with respect to stem radius, cambial age, ring width and latewood percentage in 628 the four size classes. Lines indicate class means at 15 equal intervals with respect to 629 the x-axis. For UAS and EAS n = 754 and n = 30, respectively. 630 Figure 4. Predicted tracheid widths and predicted tracheid lengths in UAS trees with 631 respect to cambial age, ring width and latewood percentage, using the models fitted to 632 data from the UAS trees. The colors indicate the width (upper panes) and length 633 (lower panes) of the tracheids with respect to the combined effects of cambial age (x- 634 axis), and ring width, or latewood percentage (y-axis). 635 Figure 5. Prediction errors of the tracheid dimensions in even-aged stands (EAS), 636 when EAS data are used as inputs to the models fitted to uneven-aged data. The 637 prediction errors are given with respect to the predicted values and the fixed variables 638 stem radius, ring width, and latewood percentages in EAS. Page 28 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Table 1. List of main variables studied in this study, their abbreviations, explanations and units. Abbreviation Explanation Unit Width Radial tracheid width µm WidthLW Radial tracheid width in latewood µm WidthEW Radial tracheid width in earlywood µm WidthTW Radial tracheid width in transitionwood µm Length Tracheid length mm R Stem radius at the position of a specific sampling point, mm CA Cambial age a RW Annual ring width mm EWW Earlywood width mm LWW Latewood width mm TWW Transitionwood width mm LWP Latewood percentage % TWP Transitionwood percentage % dbh Diameter-at-breast-height mm dbhdom Mean d of 100 largest trees per hectare mm Site The locality of studied stand (as below) dummy - LAP Lapinjärvi - VES Vesijako - SUO Suonenjoki Page 29 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Table 2. Averages, standard deviations and minimum and maximum values for radial tracheid width and tracheid length by tree size classes. Radial tracheid width (µm) UAS EAS Size class (cm) mean sd min max mean sd min max 0 – 9.99 24.84 3.15 15.42 33.10 29.63 1.70 24.90 32.83 10 – 19.99 27.58 3.47 17.59 35.79 30.67 1.87 25.63 35.95 20 – 29.99 29.21 3.89 17.10 38.26 30.54 1.65 23.49 34.75 30 – 31.67 3.85 18.92 40.43 30.48 1.89 24.45 35.94 Tracheid length (mm) UAS EAS Size class mean sd min max mean sd min max 0 – 9.99 2.41 0.54 1.08 3.32 NA NA NA NA 10 – 19.99 2.65 0.61 1.15 3.80 NA NA NA NA 20 – 29.99 2.80 0.62 1.15 4.23 3.44 0.67 2.18 4.55 30 – 3.04 0.61 1.02 4.20 3.21 0.59 2.19 3.95 Page 30 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Table 3. Parameter estimates for the model of radial tracheid width (Eq. 2), and the standard errors (SE) and statistical significances (p). For abbreviations see Table 1. Parameter Estimate SE p Fixed effects Intercept, b0 3.0390 0.0158 <0.01 Sitei, b1 LAP -0.0710 0.0123 <0.01 VES 0.0033 0.0112 0.77 SUO 0.0000 0.0000 . lnR ijk, b2 0.1194 0.0019 <0.01 CA2ijkl, b3 -0.00000177 0.000000883 0.05 RWijkl, b4 0.0487 0.0027 <0.01 RW2 ijkl, b5 -0.0053 0.0005 <0.01 ln LWPijkl, b6 -0.0331 0.0012 <0.01 ln TWPijkl, b7 -0.0253 0.0014 <0.01 Random effects Plot, σ2ij 0.0001 0.0001 0.61 Tree, σ2ijk 0.0017 0.0003 <0.01 Page 31 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Year, σ2yearl 0.0002 0.0000 <0.01 Year AR(1), ρ 0.5019 0.0139 <0.01 Residual, σ2ijkl 0.0017 0.0000 <0.01 Page 32 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Table 4. Parameter estimates for the model of tracheid length (Eq. 3), and the standard errors (SE) and statistical significances (p). For abbreviations see Table 1. Parameter Estimate SE p Fixed effects Intercept, b0 -0.0824 0.0600 0.14 Sitei, b1 LAP -0.1747 0.0263 <0.01 VES 0.0000 0.0000 . lnR ijk, b2 0.2016 0.0210 <0.01 lnCAijkl, b3 0.1698 0.0278 <0.01 CA2ijkl, b4 -0.00003060785 0.000005165204 <0.01 RWijkl, b5 -0.0184 0.0045 <0.01 ln LWPijkl, b6 -0.0156 0.0071 0.03 ln TWPijkl, b7 -0.0212 0.0099 0.03 Random effects Plot, σ2ij 0.0004 0.0007 0.59 Tree, σ2ijk 0.0026 0.0017 0.13 Year, σ2yearl 0.0001 0.0001 0.34 Page 33 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Year AR(1), ρ 0.9353 0.0133 <0.01 Residual, σ2ijkl 0.0012 0.0020 <0.01 Page 34 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Figure 1. Observed ring widths and latewood and transition wood percentages in uneven-aged and even- aged stands (UAS and EAS) with respect to cambial age in the four size classes. Lines indicate class means at 15 equal intervals with respect to the x-axis. The data correspond to those used in the measurement of the tracheid widths: for UAS and EAS n = 6004 and n = 1191 respectively. 1089x769mm (96 x 96 DPI) Page 35 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Figure 2. Observed tracheid widths in uneven-aged and even-aged stands (UAS and EAS) with respect to stem radius, cambial age, ring width and latewood percentage in the four size classes. Lines indicate class means at 15 equal intervals with respect to the x-axis. For UAS and EAS, n = 6004 and n = 1191, respectively. 176x236mm (300 x 300 DPI) Page 36 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Figure 3. Observed tracheid lengths in uneven-aged and even-aged stands (UAS and EAS) with respect to stem radius, cambial age, ring width and latewood percentage in the four size classes. Lines indicate class means at 15 equal intervals with respect to the x-axis. For UAS and EAS n = 754 and n = 30, respectively. 1060x1481mm (96 x 96 DPI) Page 37 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Figure 4. Predicted tracheid widths and predicted tracheid lengths in UAS trees with respect to cambial age, ring width and latewood percentage, using the models fitted to data from the UAS trees. The colors indicate the width (upper panes) and length (lower panes) of the tracheids with respect to the combined effects of cambial age (x-axis), and ring width, or latewood percentage (y-axis). 171x99mm (300 x 300 DPI) Page 38 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Figure 5. Prediction errors of the tracheid dimensions in even-aged stands (EAS), when EAS data are used as inputs to the models fitted to uneven-aged data. The prediction errors are given with respect to the predicted values and the fixed variables stem radius, ring width, and latewood percentages in EAS. 236x118mm (300 x 300 DPI) Page 39 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Table A1. Pearson correlation coefficients (r) and significance levels (p) between key variables in the data used for modeling tracheid width. Widthijkl WidthEWijkl WidthTWijkl WidthLWijkl Rring ijkl CAijkl RWijkl LWPijkl TWPijkl WidthEWijkl 0.962 <0.001 WidthTWijkl 0.881 0.825 <0.001 <0.001 WidthLWijkl 0.574 0.538 0.670 <0.001 <0.001 <0.001 Rring ijkl 0.765 0.769 0.707 0.503 <0.001 <0.001 <0.001 <0.001 CAijkl 0.627 0.667 0.616 0.491 0.850 <0.001 <0.001 <0.001 <0.001 <0.001 RW ijkl 0.492 0.425 0.398 0.245 0.364 0.096 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 LWPijkl -0.445 -0.346 -0.204 0.113 -0.181 0.008 -0.497 <0.001 <0.001 <0.001 <0.001 <0.001 0.527 <0.001 TWPijkl -0.125 -0.092 0.064 0.082 -0.056 -0.077 0.203 -0.042 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 dbhijk/dbhdom_ij 0.588 0.554 0.493 0.292 0.584 0.317 0.499 -0.340 -0.049 Page 40 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 where Widthijkl = Average radial tracheid width in annual ring l in tree k on plot j at site i, µm WidthEWijkl = Average radial tracheid width in latewood of annual ring l in tree k on plot j at site i, µm WidthTWijkl = Average radial tracheid width in transition wood in annual ring l in tree k on plot j at site i, µm WidthLWijkl = Average radial tracheid width in latewood in annual ring l in tree k on plot j at site i, µm Rring ijkl = Distance from pith of annual ring l in tree k on plot j at site i, mm CAijkl = Cambial age of annual ring l in tree k on plot j at site i, years RWijkl = Width of annual ring l in tree k on plot j at site i, mm LWPijkl = Latewood proportion of annual ring l in tree k on plot j at site i, % TWPijkl = Transition wood proportion of annual ring l in tree k on plot j at site i, % dijk/ddom_ij = Relative tree diameter (stem diameter of tree k on plot j at site i divided by the average diameter of the 100 thickest trees on plot j at site i) Page 41 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . i ,j, k, l = indicators of hierarchy in the data: annual ring l in tree k on plot j at site i Page 42 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . Table A2. Pearson correlation coefficients (r) and significance levels (p) between key variables in the data used for modeling tracheid length. Lengthijkm Widthijkm WidthEWijkm WidthLWijkm Rsample ijkm CAijkl RWijkm LWPijkm TWPijkm Widthijkm 0.816 <0.001 WidthEWijkm 0.833 0.980 <0.001 <0.001 WidthLWijkm 0.637 0.702 0.676 <0.001 <0.001 <0.001 Rsample ijkm 0.754 0.784 0.796 0.611 <0.001 <0.001 <0.001 <0.001 CAijkl 0.713 0.621 0.667 0.538 0.837 <0.001 <0.001 <0.001 <0.001 <0.001 RWijkm 0.189 0.461 0.400 0.286 0.341 0.025 <0.001 <0.001 <0.001 <0.001 <0.001 0.498 LWPijkm -0.252 -0.485 -0.409 -0.005 -0.194 0.025 -0.563 <0.001 <0.001 <0.001 0.897 <0.001 0.495 <0.001 TWPijkm -0.205 -0.222 -0.206 0.013 -0.127 -0.193 0.200 0.097 <0.001 <0.001 <0.001 0.719 0.001 <0.001 <0.001 0.009 dbhijk/dbhdom_ij 0.369 0.534 0.508 0.365 0.511 0.241 0.438 -0.323 -0.020 Page 43 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.590 where Lengthijkm = Average tracheid length in the 1-cm thick sample m in tree k on plot j at site i, mm Widthijkm = Average tracheid radial width in the 1-cm thick sample m in tree k on plot j at site i, mm WidthEWijkm = Average tracheid radial width in earlywood in the 1-cm thick sample m in tree k on plot j at site i, mm WidthLWijkm = Average tracheid radial width in latewood in the 1-cm thick sample m in tree k on plot j at site i, mm Rsample ijkm = Distance from pith (of the edge furthest from the pith) of the 1-cm thick sample m in tree k on plot j at site i, mm CAijkl = Cambial age of the annual ring l that lies nearest to the pith in the 1-cm thick sample m in tree k on plot j at site i, years RWijkm = Average width of annual rings in the 1-cm thick sample m in tree k on plot j at site i, mm LWPijkm = Average latewood proportion in annual rings in the 1-cm thick sample m in tree k on plot j at site i, % TWPijkm = Average transition wood proportion in annual rings in the 1-cm thick sample m in tree k on plot j at site i, % dbhijk/dbhdom_ij = Relative tree diameter (stem diameter of tree k on plot j at site i divided by the average diameter of the 100 thickest trees on plot j at site i) Page 44 of 45Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd . i, j, k, l or m = indicators of hierarchy in the data: sample m or annual ring l in tree k on plot j at site i Page 45 of 45 Canadian Journal of Forest Research (Author?s Accepted Manuscript) © The Author(s) or their Institution(s) Ca n. J. F or . R es . D ow nl oa de d fro m c dn sc ie nc ep ub .c om b y M ET LA /L EH TI SA LI o n 12 /0 8/ 21 Fo r p er so na l u se o nl y. T hi s J us t-I N m an us cr ip t i s t he a cc ep te d m an us cr ip t p rio r t o co py e di tin g an d pa ge c om po sit io n. It m ay d iff er fr om th e fin al o ffi ci al v er sio n of re co rd .