Managing plantation density through initial spacing and commercial thinning: yield results from a 60-year-old red pine spacing trial experiment

We report on a 60-year-old red pine (Pinus resinosa Aiton) spacing trial experiment located in Ontario, Canada, that included the combinations between six initial spacings (from 1.2 to 3.0 m) and the presence or absence of a commercial thinning (CT) regime, as well as their impacts on quadratic mean diameter (QMD) and stand volume yield. The CT regime, initiated at age 30, targeted a residual basal area (BA) of 38 m2·ha−1 after each of four entries. Without thinning, as initial spacing increased, QMD increased; gross and net volume production peaked in the 2.1–2.4 m spacings. With thinning, similar trends with spacing were evident for QMD, although piece sizes were larger and differences between spacings were lower. The immediate increase of mean tree size caused by tree selection explained most of the differences in QMD between thinned and unthinned plots. Thinning to a common target BA resulted in similar standing volume across spacings. Cumulative gross yield was similar between spacings of <2.1 m for both thinned and unthinned stands and decreased for thinned plots for wider spacings. Greater net volume production in thinned stands with the narrower spacings confirmed that mortality was captured. Lower gross and net production for wider spacings suggested that thinning resulted in underutilized growing space.

D r a f t D r a f t 2 26 Abstract 27 We report on a 60-year-old Pinus resinosa spacing trial experiment located in Ontario 28 (Canada) that included the combinations between six initial spacings (from 1.2-3.0 m) and 29 the presence/absence of a commercial thinning (CT) regime, and their impacts on quadratic 30 mean diameter (QMD) and stand volume yield. The CT regime, initiated at age 30, targeted a 31 residual basal area (BA) of 38 m 2 ha -1 after each of four entries. Without thinning, as initial 32 spacing increased, QMD increased; gross and net volume production peaked in the 2.1-2.4 m 33 spacings. With thinning, similar trends with spacing were evident for QMD, although piece 34 sizes were larger and differences between spacing were lower. The immediate increase of 35 average tree size caused by tree selection explained most of the differences in QMD between 36 thinned and unthinned plots. Thinning to a common target BA resulted in similar standing 37 volume across spacings. Cumulative gross yield was similar between spacings <2.1 m for 38 both unthinned/thinned stands and decreased for thinned plots for wider spacings. Greater net 39 volume production in thinned stands with the narrower spacings confirmed that mortality was 40 captured. Lower gross and net production for wider spacings suggested that thinning resulted 41 in under-utilized growing space.
D r a f t 46 Introduction D r a f t 4 71 individual tree size is maximized at lower density because of reduced intra-specific 72 competition. This pattern of an antagonistic effect of planting density on stand wood yields 73 and mean tree size has been observed for many species and many parts of the world (see 74 West 2014). Stand density can also be managed later in the development of a stand through 75 thinning operations. Stand thinning, either before or after trees have reached a commercial 76 size, is used to concentrate the finite site environmental resources on a few, selected 77 individual crop trees (Savill et al. 1997). By reducing stand density, thinning generally 78 reduces intra-specific competition and increases light and nutrient availability to residual 79 trees, hence improving their growth and reducing the rotation age to achieve a target tree size 80   (Rudolf 1990). It grows fairly rapidly compared with 87 most of North American tree species and has the potential for high value products, 88 particularly utility poles (Gilmore and Palik 2006 (Baldwin Jr. et al. 2000). Optimum planting densities and thinning intensities 98 can be derived from density management diagrams (Smith and Woods 1997), but they are 99 generally based on assumptions derived from the observations of unmanaged stands, rather 100 than empirical data from real stand-density management treatments. Long-term studies 101 investigating the main and interacting effects of initial density and thinning are highly D r a f t 6 120 quadratic mean diameter (QMD), would be greater in thinned than in unthinned plots, but that 121 these differences would decrease as initial spacing increased, because:

122
(a) the immediate increase of the average tree size due to tree selection (also known as 123 the "chainsaw effect") is more important in denser plots with such a thinning 124 prescription; 125 (b) the average release of the competition around the residual trees will be more 126 pronounced in denser plots (i.e. greater reduction of intraspecific competition), 127 because the initial competition pressure and the number of released trees per hectare 128 are greater than with plots with lower initial density.
129 As a corollary, we also expected that individual tree size would increase with increasing 130 initial spacing, but that this effect after 60 years would be less important in thinned relative to 131 unthinned plots. At the stand level, after Langsaeter's hypothesis ), we 132 expected all treatments to produce a similar cumulative gross yield, and that the thinning 133 regime would increase net production by decreasing tree mortality.  The tree selection rules applied at each thinning caused an instantaneous increase in QMD, 241 which once cumulated for the four treatments, ranged from 3.1 to 6.3 cm, in favour of the 242 narrowest spacing (Fig. 3).

243
244 Tree level response 245 Initial planting density and commercial thinning interacted to influence QMD (Table 1). In 246 unthinned plots, mean QMD did not differ significantly among initial spacings of 1.2, 1.5 and 247 1.8 m (Fig. 4). Without a CT regime to reduce density, a spacing of at least 2.1 m was 248 necessary to detect a significant increase in tree diameter compared with the narrower 249 spacings. A 9 cm gain in QMD was obtained between the 2.1 and 3 m spacings over 60 years.
250 Without thinning, a 2.4 m initial spacing was required to reach the minimal diameter 251 observed in thinned plots, which occurred in the 1.5 m spacing (Fig. 4).

252
253 Thinning slightly increased the number of large trees for the narrower spacings, but not much 254 for the larger ones. In these cases, the thinning regime eliminated the smaller size classes 4 .
255 Thinning caused a significant increase in QMD for all initial spacings; however, differences 256 among the levels of initial spacing were statistically different only when reaching the 2.4 m 257 initial spacing. The difference between thinned and unthinned plots ranged from 4.9 to 9.1 258 cm, with a correlation of -0.78 with initial spacing. However, most of this difference was 259 caused by the chainsaw effect, which was highly correlated with initial spacing (r = -0.87).
260 By subtracting the chainsaw effect from the observed difference between thinned and 261 unthinned plot, we approximated the effect of thinning on tree growth. This ranged between 262 1.9 and 3.9 cm, with no significant correlation with the initial spacing (r = -0.006).  Table 2). Merchantable volume in thinned stands was consistent across spacing 272 treatments (Table 2).

273
274 Thinning effectively reduced mortality: on average, 15.0 m 3 ha -1 across the initial spacings 275 were lost to mortality between 1982 and 2013 (Table 2). Conversely, mortality volume in the 276 controls was as high as 247 m 3 ha -1 in the 1.2 m spacing, declined as spacing increased, and 277 became relatively close to levels in thinned plots for the 2.4 and 3.0 m spacings (Table 2).

278
279 The cumulative harvest volume from the four thinnings only differed significantly between 280 the 1.2 and 3.0 m spacings, with 45% more volume harvested in narrower spacings (Table 2). 281 Cumulative gross yield was similar between thinned and unthinned plots across the range of 282 initial spacings, with the exception of the 2.1 m treatment for which the yield in unthinned 283 plots was over 180 m 3 ha -1 greater than in thinned plots, thus triggering a significant 284 interaction effect. The cumulative net yield was 22-40% greater for thinned plots for spacing 285 lower than 2.1 m, and 10-15% lower beyond this threshold (Fig. 5).

301
302 We expected that the positive influence of thinning on tree size would be dependent upon 303 initial spacing. This prediction also proved true, with a significant interaction between initial 304 planting density and the thinning treatment in driving quadratic mean diameter. We observed 305 smaller differences in tree QMD between the thinned and unthinned plots in the larger than in 306 the narrower initial spacings (e.g. a 44% increase in QMD in the 1.2 m spacing, compared 307 with a 17% increase in QMD in the 3.0 m spacing). The chainsaw effect was responsible for 308 60-70% of the differences in QMD between thinned and unthinned plots at 60 years, with a 309 strong correlation with initial spacing.  (Kozlowski 1992) and thus, sensitive to small variation in resource levels 321 (Lanner 1985). Hence, both the increase in initial spacing and the CT regime brought the 322 trees closer to optimal conditions for diameter growth, thereby reducing the relative gain 323 associated with the other treatment. Trees growing in narrower spacings were initially 324 subjected to greater levels of intraspecific competition than trees growing in wider spacings 325 (Newton 2015a). In principle, the CT regime increased the base level of environmental 326 resources, reducing the relative gain associated with a reduction in initial density. Our results 327 suggest that the thinning regime canceled the differences in resource availability among 328 initial spacing plots: once the chainsaw effect was removed from the observed difference in 329 QMD (thus estimating the difference in QMD caused by growth and survival after thinning), 330 QMD was still greater for thinned plots in comparison with unthinned ones, however with 331 little significant difference between initial spacings. 332 333 At the stand level, we expected all treatments to produce a similar cumulative gross yield 334 , and that the thinning regime would have captured tree mortality.
335 However, we observed that cumulative gross yield varied as a function of interactions 336 between the initial planting density and the CT regime, and that some treatment combinations 337 led to significantly greater cumulative gross yield than others. The trend towards greater