Assessing the effects of thinning on stem growth allocation of individual Scots pine trees

Forest management alters the growing conditions and thus further development of trees. However, quantitative assessment of forest management on tree growth has been demanding as methodologies for capturing changes comprehensively in space and time have been lacking. Terrestrial laser scanning (TLS) has shown to be capable of providing three-dimensional (3D) tree stem reconstructions required for revealing differences between stem shapes and sizes. In this study, we used 3D reconstructions of tree stems from TLS and an unmanned aerial vehicle (UAV) to investigate how varying thinning treatments and the following growth effects affected stem shape and size of Scots pine (Pinus sylvestris L.) trees. The results showed that intensive thinning resulted in more stem volume and therefore total biomass allocation and carbon uptake compared to the moderate thinning. Relationship between tree height and diameter at breast height (i.e. slenderness) varied between both thinning intensity and type (i.e. from below and above) indicating differing response to thinning and allocation of stem growth of Scots pine trees. Furthermore, intensive thinning, especially from below, produced less variation in relative stem attributes characterizing stem shape and size. Thus, it can be concluded that thinning intensity, type, and the following growth effects have an impact on post-thinning stem shape and size of Scots pine trees. Our study presented detailed measurements on post-thinning stem growth of Scots pines that have been laborious or impracticable before the emergence of detailed 3D technologies. Moreover, the stem reconstructions from TLS and UAV provided variety of attributes characterizing stem shape and size that have not traditionally been feasible to obtain. The study demonstrated that detailed 3D technologies, such as TLS and UAV, provide information that can be used to generate new knowledge for supporting forest management and silviculture as well as improving ecological understanding of boreal forests.

used 3D reconstructions of tree stems from TLS and an unmanned aerial vehicle (UAV) to investigate 23 how varying thinning treatments and the following growth effects affected stem shape and size of 24 Scots pine (Pinus sylvestris L.) trees. The results showed that intensive thinning resulted in more stem 25 volume and therefore total biomass allocation and carbon uptake compared to the moderate thinning. 26 Relationship between tree height and diameter at breast height (i.e. slenderness) varied between both 27 thinning intensity and type (i.e. from below and above) indicating differing response to thinning and 28 allocation of stem growth of Scots pine trees. Furthermore, intensive thinning, especially from below, 29 produced less variation in relative stem attributes characterizing stem shape and size. Thus, it can be 30 concluded that thinning intensity, type, and the following growth effects have an impact on post-31 thinning stem shape and size of Scots pine trees. Our study presented detailed measurements on post-32 thinning stem growth of Scots pines that have been laborious or impracticable before the emergence 33 of detailed 3D technologies. Moreover, the stem reconstructions from TLS and UAV provided variety 34 of attributes characterizing stem shape and size that have not traditionally been feasible to obtain. The 35 study demonstrated that detailed 3D technologies, such as TLS and UAV, provide information that 36 can be used to generate new knowledge for supporting forest management and silviculture as well as 37 improving ecological understanding of boreal forests.

43
Allocation of photosynthesis products between different parts of a tree define their development. If 44 resources are limited, trees first channel them towards respiration and sustaining existing parts, 45 producing fine roots and seed (i.e. reproduction), height growth and only then growth in diameter at 46 breast height (1.3 m, DBH). Competition for resources (e.g. light, water, nutrients) between trees also 47 affects development of different tree parts. Thus, differences in the allocation of photosynthesis 48 products within a tree and competition between trees result in variation in tree architecture. 49 50 In thinning, as a part of forest management, competition within a population is regulated as part of 51 the trees are removed. Therefore, ecologically thinning is aimed at improving growing conditions 52 (i.e., light, temperature, water, nutrients) of remaining trees and economically to maximize the net 53 present value of a stand by decreasing the opportunity costs of the capital. Especially when part of 54 shadowing leaf mass is removed in thinning, the amount of light is increased, and growth of remaining 55 trees is enhanced (White 1980). Additionally, an increase in the amount of nutrients, especially in  Intermediate and suppressed trees can expect to benefit from thinning more than co-dominant and 63 dominant trees as growing conditions are improved and competition decreased (i.e. more available 64 light) for trees in lower canopy layers. Mäkinen & Isomäki (2004c) reported that relative growth of intensities, even the heavy thinning, studied in Mäkinen & Isomäki (2004c) correspond 96 approximately thinning intensities applied nowadays in Finland (Rantala 2011). whereas thinning from below increased growth of dominant height more than thinning from above 104 (Vuokila 1977). Similarly, Mielikäinen & Valkonen (1991) reported that thinning from above 105 resulted in 4-8% more volume growth in Scot pine stands than thinning from below. Furthermore, 106 Eriksson & Karlsson (1997) illustrated that thinning from above did not decrease volume growth in 107 Scots pine stands in comparison with thinning from below.

109
Growing conditions are, thus, dependent on tree density that affects growth and structure of trees 110 (Harper 1977) whereas changes in height and diameter growth alter stem shape. Therefore, 111 understanding changes in stem shape enable assessing the effects of thinning on future growth as well    Although theoretical understanding exists on the effects of forest management and resulting growing 152 conditions on allocation between height and diameter growth of trees, understanding differences in 153 tree architecture based on quantitative information is still limited. However, detailed information on 154 stem shape and size has not been available for studies investigating effects of both thinning intensity 155 and type. Therefore, the objective of this study is to investigate the effects of intensity and type of 156 thinning on post-thinning stem growth of Scots pine trees. Point clouds from TLS and UAV provid 157 an opportunity in generating various new attributes characterizing both absolute and relative stem 158 shape and size. Based on previous studies on growth and yield, we hypothesized that thinning 159 intensity and type result in differing stem shape and size. It was further divided to following research 160 questions: (1) how thinning intensity and type affect stem shape and size; (2) how thinning intensity 161 and type affect growth allocation between DBH and tree height; and (3) how thinning intensity and 162 type affect variation in stem shape and size. Vesijako, respectively. Also, first thinning had been carried out in early 1990s for all study sites. The experimental design of the study sites includes two levels of thinning intensity and three thinning 183 types resulting in six different thinning treatments, namely i) moderate thinning from below, ii) 184 moderate thinning from above, iii) moderate systematic thinning from above, iv) intensive thinning 185 from below, v) intensive thinning from above, and vi) intensive systematic thinning from above, as 186 well as a control plot where no thinning has been carried out since the establishment. Moderate   Tree species, DBH from two perpendicular directions, crown layer, and health status were recorded 204 from each tree within a plot for each measurement time. The proportion of Norway spruce and 205 deciduous trees (i.e. Betula sp and Alnus sp) from the total stem volume of all trees within the 27 206 sample plots was 3.06% and 0.03%, respectively. Each sample plot also includes 22 sample trees, on 207 average, from which tree height, crown base height, and height of the lowest dead branch were also 208 measured. Stand attributes before and after thinning treatments together with thinning removals are 209 presented in Table 1 and the development of tree-level attributes for each thinning treatment is found 210 in Table 2. The remaining relative stand basal area after moderate thinning was ~68% of the stocking 211 before thinning and intensive thinning reduced stocking levels down to 34%. There were no large 212 differences in remaining basal area or volume between thinning types with the same thinning intensity 213 ( USA) for all three study sites between September and October 2018. Eight scans were place to each 229 sample plot, the scan setup is depicted in Figure 2, and scan resolution of point distance approximately   In addition to TLS data, aerial images were obtained by using an UAV with Gryphon Dynamics 242 quadcopter frame. Two Sony A7R II digital cameras that had CMOS sensor of 42.4 MP, with a Sony 243 FE 35mm f/2.8 ZA Carl Zeiss Sonnar T* lens, were mounted on the UAV in +15° and -15° angles.    Table 3. Form factor at breast height (ff), tapering (taper), and slenderness (slend) have 285 been used in assessing stem shape and were also included in this study. Cumulative volume (e.g.

286
height at which 50% of stem volume accumulated (h_vol50), and volume percentiles (p10, .., p90)) 287 enable more detailed quantitative assessment of stem shape and size. Volume percentiles were 288 defined as the height at which each percentage point of stem volume was accumulated. Furthermore, 289 absolute and relative volume and tapering below and above 50% of tree height, represented attributes 290 that can be generated from a taper curve. As taper curve has been impractical to measure, these kinds   Table 3. Single tree attributes characterizing stem shape and size derived from taper curves generated 300 with information from terrestrial laser scanning and unmanned aerial vehicles.

308
where is each stem attribute described in Table 3  were estimated with a smoothing cubic spline to relative heights of 1%, 2.5%, 5%, 7.5%, 10%, 15%, 324 20%, …, 95%, and 100% using the taper curve measurements obtained with information from TLS 325 and UAV. This was carried out to normalize the differences in tree height. Finally, a mean taper curve 326 was calculated for each treatment together with its standard deviation.

328
As especially thinning type implicitly affects the size of the remaining trees (e.g. in thinning from 329 below small trees were removed), we examined the basic statistics (i.e. minimum, mean, maximum, 330 and standard deviation) of relative stem attributes, namely relative volume, relative volume below 331 and above 50% tree height as well as relative tapering below and above 50% of tree height but also 332 slenderness and both form factor at breast height and form factor up to 50% of tree height. These 333 attributes disregard tree size and demonstrate more objectively the effects of different thinning 334 intensity and type and can, therefore, reveal the variation in stem shape and size.

336
Although taper curve for each tree was only available for one time point, field measurements before 337 and after the thinning treatments were on hand. Post-thinning growth in DBH, height, and stem 338 volume was calculated for all live trees that were in the sample plots during the last measurements.

339
Additionally, difference in slenderness was assessed as a ratio between the first and last measurements Tukey's honest significance test) were utilized in testing difference in these change-related attributes 345 between the thinning treatments. Additionally, difference in ratio between stem volume from the first 346 and last field measurement was also included to disregard the tree size in growth assessments.

350
When the effects of a same thinning type (i.e. from below, from above, and systematic) but of 351 different intensity (i.e. moderate and intensive) on the stem attributes characterizing stem shape and 352 size were compared, intensive thinning mainly produced larger attributes characterizing absolute tree 353 size (Figure 3). Similarly, thinning from below (both moderate and intensive) mostly resulted in larger 354 stem attributes characterizing absolute stem size compared to respective attributes caused by thinning 355 from above and systematic with the same intensity ( Figure 3). In relative attributes that disregard tree 356 size, such as relative volumes and taperings as well as form factor at breast height, however, the 357 difference between thinning intensity and type was smaller (Figure 3). On the contrary, the difference 358 in slenderness and form factor up to 50% of tree height between thinning intensity and type was 359 visible. There was a noticeable trend between moderate thinning types in stem volume percentiles, in 360 other words in moderate systematic thinning and thinning from above stem volume percentiles were 361 accumulated at lower heights of a stem compared to moderate thinning from below ( Figure 4).

362
Between intensive thinning types there was no such trend but intensive thinning from above, among 363 the intensive thinnings, produced the lowest heights for the stem volume percentile accumulation, 364 indicating larger volume accumulation in lower heights of a stem.  Table 3 for the descriptions of the attributes. The nested two-level linear mixed-effects models provided quantitative details about differences in 384 stem attributes between thinning treatments depicted in Figures 3 and 4 but also whether their effects 385 were statistically significant (Table S1). Variances of the random part of the mixed-effects model 386 explained the variation in the stem attributes between the study sites and the sample plots within a 387 study site. They were rather similar for all other stem attributes except for form quotient for which 388 the variance of a study site was a ten of times larger (3.128 2 ) than the plot-level variance (0.004 2 ) 389 (Table S2). Furthermore, for relative volume (i.e. ratio between volume below and above 50% of tree 390 height) the plot-level variance was considerably larger (6.326 2 ) than the variance at study site level 391 (0.014 2 ) similar to tapering above 50% of tree height (plot level 1.464 2 vs study site 0.001 2 ). These 392 suggest that most of the variation between the stem attributes was explained by both a study site and 393 a plot within a study site, but a study site affected considerably more on form quotient whereas a 394 sample plot affected considerably more on relative volume and (absolute) tapering above 50% of tree 395 height.

397
As the differences were not noticeable for all stem attributes between thinning treatments in Figures   398 3 and 4, the analysis of variance only revealed statistically significant differences (p-value ≤ 0.05) in 399 DBH, volume, volume below and above 50% of tree height, relative volume, relative volume below 400 50% of tree height, height at which 50% of total stem volume accumulated, slenderness, form factor 401 up to 50% of tree height, tapering as well as tapering below and above 50% of tree height. However, 402 it did not reveal between which treatments the differences occurred. Tukey's honest significance test, 403 however, showed statistically significant difference (p-value ≤ 0.05) between thinning treatments 404 (Table S3). Furthermore, the analysis of variance showed that there was a significant difference (p-405 value ≤ 0.05) between the three study sites in all stem attributes except form quotient and relative 406 tapering above 50% of tree height. When the analysis of variance was used for assessing difference 407 between plots within a study site, there was statistically significant difference (p-value ≤ 0.05) in all 408 stem attributes except relative volume and relative volume below 50% of tree height.

410
When comparing the influence of thinning intensity on the relative stem attributes, only relative 411 volume and relative volume above 50% of tree height differed significantly (p-value ≤ 0.05) between 412 moderate thinning from above and intensive thinning from below. The thinning type within the 413 intensive thinnings affected relative volume and relative volume above 50% of tree height from the 414 relative stem attributes; they differed between intensive from below and systematic. However, 415 statistically significant difference was consistently found in slenderness when comparing both 416 thinning intensity and type (Table S3) whereas form factor up to 50% of tree height differed 417 statistically significantly between thinning types of intensive thinning. The relative volume and tapering attributes as well as slenderness and form factors were assessed 434 separately through basic statistics to examine the variation due to the thinning treatments. The relative 435 attributes do not consider tree size that is implicitly included in thinning type (i.e. tree size affects 436 which trees are removed). There were no clear differences between thinning treatments in relative 437 volumes (i.e. relative volume, relative volume below and above 50% of tree height) or taperings 438 ( Figure 3) and statistically significant difference was only found in relative volume and relative 439 volume above 50% of tree height (Tables 4 and S3). The intensive thinnings (i.e. below, above, and 440 systematic) mainly resulted in smaller standard deviation compared to the relative volume and 441 tapering attributes of their corresponding moderate counterparts (Table 4). Only between moderate 442 and intensive thinning from above the standard deviation of relative tapering above 50% of tree height 443 was very similar, namely 0.006 and 0.007, respectively. When the thinning type was compared, 444 thinning from below produced smaller standard deviation than thinning from above and systematic  (Table 4).  Growth in DBH, height, and stem volume was calculated based on the first and last field 472 measurements (Table 5). Additionally, difference in slenderness between the first and last field whereas difference in all these change-related attributes was statistically significant between plots 482 within study sites.. In DBH growth and difference in slenderness significant difference (p-value ≤ 483 0.05) was found between thinning intensity whereas significant difference (p-value ≤ 0.05) in volume 484 growth was found between thinning type (intensive thinning treatments). When ratio of stem volume 485 between the first and last field measurements was assessed, the significant difference (p-value ≤ 0.05) 486 was found between thinning intensity and type but only between intensive systematic thinning.and 487 intensive thinnings from below and above.  The objective of this work was to examine how thinning intensity but also thinning type (i.e. from 497 below, from above, and systematic) influenced stem shape and size of Scots pine trees. The results

498
showed that there was no clear difference in relative stem attributes characterizing stem shape and 499 size although in absolute stem attributes difference was found between thinning type (research 500 question 1). However, the relationship between DBH and height (i.e. slenderness) differed between 501 both thinning intensity and type (research question 2). Finally, when assessing the similarity of the

551
Our results did not reveal notable difference in form factor between moderate and intensive thinning 552 whereas tapering below 50% of tree height was smaller for moderate thinnings than for intensive Intensive thinning expanded the growing space of and reduced the competition between the remaining 575 trees allowing them to allocate the photosynthesis products not just sustaining respiration and existing 576 parts but also for wood formation and growth.