Effect of thinning on the relationship between mean ring density and climate in black spruce (Picea mariana (Mill.)B.S.P.)

Effect of thinning on the relationship between mean ring density and climate in black spruce... Abstract Relationships between wood density and climatic variables have generally been developed from unmanaged stands near the treeline. Using black spruce (Picea mariana (Mill.) B.S.P.) samples from the managed boreal zone in Canada, we investigated whether the relationship between mean ring density (MRD) and climatic variables is altered by silvicultural practices. We analysed the MRD of 10 384 growth rings from 72 trees sampled among 18 stands (nine thinned, nine controls) across Quebec, Canada. We constructed a mixed-effects model relating MRD to cambial age and ring width (RW). Model residuals (εM1), i.e. the difference between observed and predicted MRD, were then related to monthly temperature and precipitation of the year of ring formation and the year before. After thinning, RW slightly increased while MRD remained constant, thus lowering the strength of the relationship between MRD and RW. εM1 were positively related to spring temperatures and negatively related to summer temperatures and precipitation. No effect of thinning on the relationship between εM1 and climatic variables was observed. The sample trees grew in less limiting conditions than at the treeline so the reduced strength of the relationship between MRD and growth rate in thinned stands may result from a higher photosynthetic capacity. Such results may have implications in forest management as thinning could increase the value of black spruce wood. Introduction Conifer wood density is influenced by a large number of variables. Among them, some result from internal causes, such as ageing or growth rate (Zobel and Buijtenen, 1989), while others originate from external causes, such as climate or competition (Zobel and Buijtenen, 1989; Fritts et al., 1991). For several conifer species, mean ring density (MRD) decreases abruptly from the pith to the juvenile wood, after which it slowly increases in the mature wood (Lachenbruch et al., 2011; Xiang et al., 2014b). These changes are often associated with differences in functional requirements (Lachenbruch et al., 2011): the juvenile wood needs to be flexible but strong, while mature wood needs to be mechanically stiff and ensure high hydraulic conductivity. The radial growth rate (referred hereafter as ‘ring width’) of the stem can also be negatively correlated with wood density, especially in the Picea genus (Macdonald and Hubert, 2002; Xiang et al., 2014b). This is often related to changes in the ring structure: with a higher growth rate, the proportion of dense latewood tends to decrease, which in turn affects MRD (Brazier, 1970; Schneider et al., 2008; Franceschini et al., 2013b). Silvicultural treatments may also influence MRD in conifers (Zobel and Buijtenen, 1989). As a general rule, thinning results in an increase in the rate of volume growth (Goudiaby et al., 2012; Soucy et al., 2012) and radial growth of individual trees (Vincent et al., 2009), which leads to a decrease in wood density (Jaakkola et al., 2005). However, unexplained deviations from this general trend were observed in pine species (Peltola et al., 2007; Schneider et al., 2008) and in Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco, Todaro and Macchioni, 2010). In addition, few studies have observed a decrease in wood density following thinning in spruce species in boreal conditions and when observed, the decrease was generally moderate (Picea abies (L.) Karst., Jaakkola et al., 2005) or non-significant (Picea mariana (Mill.) B.S.P., Tong et al., 2011; Vincent et al., 2011). The effects of climate on wood density have been widely studied, especially in the context of past climatic reconstruction (Wang et al., 2001; Vaganov et al., 2011). Density-based dendroclimatology generally uses the maximum ring density of conifers growing at the altitudinal or latitudinal treeline (Hughes, 2002), because the dense latewood is formed during the summer when the climatic constraints are the strongest (Wang et al., 2002; Deslauriers et al., 2003). However, studies performed on spruce species growing in less limiting conditions (i.e. far from the treeline) also suggested that both latewood density (Franceschini et al., 2013a) and MRD can be under significant climatic determinism (Wimmer and Grabner, 2000; Xiang et al., 2014a). When detailing these effects of climatic variables on wood density, it can be derived that high summer temperatures are generally associated with high latewood proportions, and thus high MRD (Wimmer and Grabner, 2000; Franceschini et al., 2013a). Previous studies also reported negative relationships between MRD and summer precipitation (Wang et al., 2002; Xiang et al., 2014a). The effects of silviculture and climatic variables on MRD have been the subject of several studies. Some studies aimed to model simultaneously the effects of thinning and climatic variables on wood density in Douglas fir (Kantavichai et al., 2010; Filipescu et al., 2013), but potential interactions were not investigated. After thinning, growth conditions of residual trees are generally characterized by increased light (Hale, 2003; Ma et al., 2010) and water availability (Aussenac and Granier, 1988; Gebhardt et al., 2014). Therefore, thinning could modify the relationship between MRD and climatic variables. For example, thinning is known to increase diameter growth under drought conditions compared to unthinned stands in Norway spruce (P. abies (L.) Karst., Kohler et al., 2010; Misson et al., 2003), especially in young stands (D’Amato et al., 2013). In addition, conifers have developed mechanisms to reduce the risk of drought-induced cavitation (i.e. embolism), which consists mainly of producing cells with small lumens (Hacke et al., 2001), thus leading to the production of high-density wood (Hacke and Sperry, 2001). This study investigates the effect of thinning on the relationship between climatic variables and mean ring density in black spruce (P. mariana (Mill.) B.S.P.). This species is both widely distributed in North America (Burns and Honkala, 1990) and has considerable economic importance (Bustos et al., 2003). In Quebec (Canada), commercial thinning has been suggested for these stands as a means to modify stand structure in the ecosystemic management strategy (Gauthier and Vaillancourt, 2008) or to promote intensive wood production (Soucy et al., 2012; Ministère des Ressources Naturelles, 2015). The objectives of our study were thus twofold: (i) to identify the effect of thinning on MRD and (ii) to determine the effects of thinning on the relationship between wood density and climatic variables. We chose to work on MRD, because it is an important indicator of wood quality for black spruce (Kennedy, 1995; Vincent et al., 2011) and an influential variable in wood biomass computations (Bouriaud et al., 2005). We first hypothesized that the strength of the relationship between MRD and growth rate decreases after thinning. Our second hypothesis was that thinning also reduces the strength of the relationship between MRD and climatic variables. Materials and methods Stand characteristics and sampling design We collected data from three regions of Quebec, i.e. from West to East, Abitibi, Lake Saint-John and North Shore (Figure 1). The distance between the sampling locations of Abitibi and North Shore was almost 1000 km. In each region, three stands were selected from a large provincial network of permanent plots. The thinning trials were established by the Ministry of Forests, Wildlife and Parks of Quebec (MFWPQ) to investigate long-term effects of operational commercial thinning on stand development. A total of nine stands (three per site) were selected according to three criteria: (i) the proportion of black spruce trees had to be higher than 75 per cent of the total basal area, (ii) tree age at the year of the thinning had to be between 50 and 60 years and (iii) sites had to be accessible by road. The thinning treatment was a commercial thinning from below, which targeted the removal of the least vigorous trees from dense even-aged stands. In each stand, 400-m2 circular sample plots had been established by the MFWPQ in the treated area. A control plot was also established within a 1-ha untreated block in the vicinity of the thinned plot. The characteristics of the sample stands are described in Table 1. Table 1 Stand and tree characteristics in 2010. Region Site Treatment Longitude (degrees) Latitude (degrees) Tree age DBH (cm) Tree height (m) North-Shore 1 Control stand −69.4511 49.1501 94 (23) 13.9 (1.6) 12.9 (2.0) Thinned stand −69.4510 49.1507 84 (10) 13.9 (3.4) 18.0 (0.3) 2 Control stand −69.4533 49.1520 85 (2) 12.6 (2.4) 18.3 (1.9) Thinned stand −69.4534 49.1525 80 (6) 17.9 (2.9) 15.3 (2.3) 3 Control stand −69.4327 49.1520 70 (13) 11.0 (1.1) 13.6 (1.1) Thinned stand −69.4332 49.1520 82 (7) 13.7 (2.7) 14.5 (0.5) Lake-Saint-John 4 Control stand −70.4740 49.2435 82 (28) 14.7 (1.7) 10.3 (1.8) Thinned stand −70.4740 49.2441 72 (8) 17.0 (1.9) 12.3 (2.8) 5 Control stand −70.4718 49.2409 79 (5) 17.2 (5.6) 11.0 (2.1) Thinned stand −70.4711 49.2408 78 (6) 22.9 (1.5) 13.7 (3.1) 6 Control stand −72.1824 49.1047 62 (5) 15.0 (2.9) 8.9 (1.4) Thinned stand −72.1824 49.1047 66 (3) 17.9 (0.7) 12.5 (1.7) Abitibi 7 Control stand −77.0705 48.2038 65 (6) 14.4 (2.3) 13.5 (2.7) Thinned stand −77.0716 48.2035 73 (7) 17.2 (1.6) 13.1 (1.5) 8 Control stand −77.0459 48.1848 82 (4) 17.8 (1.3) 13.6 (4.3) Thinned stand −77.0447 48.1853 82 (6) 17.3 (2.4) 17.3 (1.3) 9 Control stand −77.5142 48.4831 91 (5) 14.0 (1.0) 16.1 (2.1) Thinned stand −77.5146 48.4825 88 (9) 14.1 (2.0) 16.8 (1.4) Region Site Treatment Longitude (degrees) Latitude (degrees) Tree age DBH (cm) Tree height (m) North-Shore 1 Control stand −69.4511 49.1501 94 (23) 13.9 (1.6) 12.9 (2.0) Thinned stand −69.4510 49.1507 84 (10) 13.9 (3.4) 18.0 (0.3) 2 Control stand −69.4533 49.1520 85 (2) 12.6 (2.4) 18.3 (1.9) Thinned stand −69.4534 49.1525 80 (6) 17.9 (2.9) 15.3 (2.3) 3 Control stand −69.4327 49.1520 70 (13) 11.0 (1.1) 13.6 (1.1) Thinned stand −69.4332 49.1520 82 (7) 13.7 (2.7) 14.5 (0.5) Lake-Saint-John 4 Control stand −70.4740 49.2435 82 (28) 14.7 (1.7) 10.3 (1.8) Thinned stand −70.4740 49.2441 72 (8) 17.0 (1.9) 12.3 (2.8) 5 Control stand −70.4718 49.2409 79 (5) 17.2 (5.6) 11.0 (2.1) Thinned stand −70.4711 49.2408 78 (6) 22.9 (1.5) 13.7 (3.1) 6 Control stand −72.1824 49.1047 62 (5) 15.0 (2.9) 8.9 (1.4) Thinned stand −72.1824 49.1047 66 (3) 17.9 (0.7) 12.5 (1.7) Abitibi 7 Control stand −77.0705 48.2038 65 (6) 14.4 (2.3) 13.5 (2.7) Thinned stand −77.0716 48.2035 73 (7) 17.2 (1.6) 13.1 (1.5) 8 Control stand −77.0459 48.1848 82 (4) 17.8 (1.3) 13.6 (4.3) Thinned stand −77.0447 48.1853 82 (6) 17.3 (2.4) 17.3 (1.3) 9 Control stand −77.5142 48.4831 91 (5) 14.0 (1.0) 16.1 (2.1) Thinned stand −77.5146 48.4825 88 (9) 14.1 (2.0) 16.8 (1.4) Standard deviations are given in parentheses. Table 1 Stand and tree characteristics in 2010. Region Site Treatment Longitude (degrees) Latitude (degrees) Tree age DBH (cm) Tree height (m) North-Shore 1 Control stand −69.4511 49.1501 94 (23) 13.9 (1.6) 12.9 (2.0) Thinned stand −69.4510 49.1507 84 (10) 13.9 (3.4) 18.0 (0.3) 2 Control stand −69.4533 49.1520 85 (2) 12.6 (2.4) 18.3 (1.9) Thinned stand −69.4534 49.1525 80 (6) 17.9 (2.9) 15.3 (2.3) 3 Control stand −69.4327 49.1520 70 (13) 11.0 (1.1) 13.6 (1.1) Thinned stand −69.4332 49.1520 82 (7) 13.7 (2.7) 14.5 (0.5) Lake-Saint-John 4 Control stand −70.4740 49.2435 82 (28) 14.7 (1.7) 10.3 (1.8) Thinned stand −70.4740 49.2441 72 (8) 17.0 (1.9) 12.3 (2.8) 5 Control stand −70.4718 49.2409 79 (5) 17.2 (5.6) 11.0 (2.1) Thinned stand −70.4711 49.2408 78 (6) 22.9 (1.5) 13.7 (3.1) 6 Control stand −72.1824 49.1047 62 (5) 15.0 (2.9) 8.9 (1.4) Thinned stand −72.1824 49.1047 66 (3) 17.9 (0.7) 12.5 (1.7) Abitibi 7 Control stand −77.0705 48.2038 65 (6) 14.4 (2.3) 13.5 (2.7) Thinned stand −77.0716 48.2035 73 (7) 17.2 (1.6) 13.1 (1.5) 8 Control stand −77.0459 48.1848 82 (4) 17.8 (1.3) 13.6 (4.3) Thinned stand −77.0447 48.1853 82 (6) 17.3 (2.4) 17.3 (1.3) 9 Control stand −77.5142 48.4831 91 (5) 14.0 (1.0) 16.1 (2.1) Thinned stand −77.5146 48.4825 88 (9) 14.1 (2.0) 16.8 (1.4) Region Site Treatment Longitude (degrees) Latitude (degrees) Tree age DBH (cm) Tree height (m) North-Shore 1 Control stand −69.4511 49.1501 94 (23) 13.9 (1.6) 12.9 (2.0) Thinned stand −69.4510 49.1507 84 (10) 13.9 (3.4) 18.0 (0.3) 2 Control stand −69.4533 49.1520 85 (2) 12.6 (2.4) 18.3 (1.9) Thinned stand −69.4534 49.1525 80 (6) 17.9 (2.9) 15.3 (2.3) 3 Control stand −69.4327 49.1520 70 (13) 11.0 (1.1) 13.6 (1.1) Thinned stand −69.4332 49.1520 82 (7) 13.7 (2.7) 14.5 (0.5) Lake-Saint-John 4 Control stand −70.4740 49.2435 82 (28) 14.7 (1.7) 10.3 (1.8) Thinned stand −70.4740 49.2441 72 (8) 17.0 (1.9) 12.3 (2.8) 5 Control stand −70.4718 49.2409 79 (5) 17.2 (5.6) 11.0 (2.1) Thinned stand −70.4711 49.2408 78 (6) 22.9 (1.5) 13.7 (3.1) 6 Control stand −72.1824 49.1047 62 (5) 15.0 (2.9) 8.9 (1.4) Thinned stand −72.1824 49.1047 66 (3) 17.9 (0.7) 12.5 (1.7) Abitibi 7 Control stand −77.0705 48.2038 65 (6) 14.4 (2.3) 13.5 (2.7) Thinned stand −77.0716 48.2035 73 (7) 17.2 (1.6) 13.1 (1.5) 8 Control stand −77.0459 48.1848 82 (4) 17.8 (1.3) 13.6 (4.3) Thinned stand −77.0447 48.1853 82 (6) 17.3 (2.4) 17.3 (1.3) 9 Control stand −77.5142 48.4831 91 (5) 14.0 (1.0) 16.1 (2.1) Thinned stand −77.5146 48.4825 88 (9) 14.1 (2.0) 16.8 (1.4) Standard deviations are given in parentheses. Figure 1 View largeDownload slide Site location map. The species distribution map originates from Little (1971). Figure 1 View largeDownload slide Site location map. The species distribution map originates from Little (1971). The commercial thinning was planned to be conducted at least 15 years prior to the final harvest, with a reduction of 28–35 per cent of the initial stand basal area (MRNFP, 2003). Given variations in skid trail proportion between plots, the actual thinning intensity at the plot level ranged between 18 and 44 per cent of the initial stand basal area. Dendrometric measurements on permanent plots were made before and after harvesting. The thinning was conducted between 1997 and 2000 depending on the region and site (Table 2), while the tree sampling for the present study was conducted in 2010. In each plot, the centre of a variable-radius sampling plot was established with a factor 2 prism. Starting from a random azimuth and turning clockwise, the first, third, fifth and seventh merchantable trees (DBH > 9 cm) were selected for destructive sampling. Four trees were sampled in each of the 18 plots (3 regions × 3 sites per region × 2 plots per site) for a total of 72 black spruce trees. Table 2 Tree and stand characteristics 10 years before and after the thinning. Region Site number Treatment Year of thinning Mean ring density (kg m−3) Ring width (mm) Basal area (m2 ha−1) Stand density (stems ha−1) Before thinning After thinning Before thinning After thinning Before thinning After thinning Before thinning After thinning North-Shore 1 Control 1997 526 (75) 510 (78) 0.65 (0.20) 0.71 (0.21)* 33.7 2150 Thinned 1997 519 (62) 521 (68) 0.76 (0.32) 0.71 (0.39) 42.6 26.7 2450 1600 2 Control 1997 561 (60) 551 (73) 0.91 (0.38) 0.84 (0.42) 31.8 2975 Thinned 1997 527 (59) 509 (57)** 0.76 (0.31) 0.71 (0.35) 26.8 19.1 2675 1800 3 Control 1997 568 (71) 554 (61) 0.57 (0.21) 0.57 (0.29) 35.9 3425 Thinned 1997 582 (116) 576 (115) 0.81 (0.34) 0.80 (0.36) 38.9 33.0 2600 2050 Lake-Saint-John 4 Control 1998 504 (84) 513 (79) 0.52 (0.28) 0.62 (0.40)* 37.1 2925 Thinned 1998 506 (41) 512 (71) 1.14 (0.47) 1.01 (0.40)* 40.1 29.7 2500 1575 5 Control 1998 524 (41) 554 (44)*** 0.89 (0.41) 0.73 (0.38)*** 36.8 2550 Thinned 1998 513 (46) 513 (41) 1.08 (0.31) 1.07 (0.32) 43.4 24.3 2050 925 6 Control 1998 489 (48) 508 (55)*** 1.41 (0.49) 1.14 (0.46)*** 34.6 1675 Thinned 1998 536 (59) 512 (58)*** 0.79 (0.24) 0.97 (0.41)*** 29.0 25.0 1775 1400 Abitibi 7 Control 1999 570 (68) 587 (72) 1.02 (0.42) 0.85 (0.32)** 26.8 1175 Thinned 1999 529 (47) 539 (53)* 0.92 (0.35) 1.01 (0.33)** n.m. 21.5 n.m. 1150 8 Control 1999 511 (49) 513 (64) 0.86 (0.27) 0.86 (0.26) 36.8 1850 Thinned 1999 590 (89) 560 (73)** 0.61 (0.25) 0.72 (0.38)** 36.3 23.0 2700 1350 9 Control 2000 565 (77) 570 (62) 0.75 (0.24) 0.74 (0.28) 31.1 2275 Thinned 2000 540 (48) 544 (47) 0.84 (0.34) 0.79 (0.36) 26.0 16.6 2150 1250 Region Site number Treatment Year of thinning Mean ring density (kg m−3) Ring width (mm) Basal area (m2 ha−1) Stand density (stems ha−1) Before thinning After thinning Before thinning After thinning Before thinning After thinning Before thinning After thinning North-Shore 1 Control 1997 526 (75) 510 (78) 0.65 (0.20) 0.71 (0.21)* 33.7 2150 Thinned 1997 519 (62) 521 (68) 0.76 (0.32) 0.71 (0.39) 42.6 26.7 2450 1600 2 Control 1997 561 (60) 551 (73) 0.91 (0.38) 0.84 (0.42) 31.8 2975 Thinned 1997 527 (59) 509 (57)** 0.76 (0.31) 0.71 (0.35) 26.8 19.1 2675 1800 3 Control 1997 568 (71) 554 (61) 0.57 (0.21) 0.57 (0.29) 35.9 3425 Thinned 1997 582 (116) 576 (115) 0.81 (0.34) 0.80 (0.36) 38.9 33.0 2600 2050 Lake-Saint-John 4 Control 1998 504 (84) 513 (79) 0.52 (0.28) 0.62 (0.40)* 37.1 2925 Thinned 1998 506 (41) 512 (71) 1.14 (0.47) 1.01 (0.40)* 40.1 29.7 2500 1575 5 Control 1998 524 (41) 554 (44)*** 0.89 (0.41) 0.73 (0.38)*** 36.8 2550 Thinned 1998 513 (46) 513 (41) 1.08 (0.31) 1.07 (0.32) 43.4 24.3 2050 925 6 Control 1998 489 (48) 508 (55)*** 1.41 (0.49) 1.14 (0.46)*** 34.6 1675 Thinned 1998 536 (59) 512 (58)*** 0.79 (0.24) 0.97 (0.41)*** 29.0 25.0 1775 1400 Abitibi 7 Control 1999 570 (68) 587 (72) 1.02 (0.42) 0.85 (0.32)** 26.8 1175 Thinned 1999 529 (47) 539 (53)* 0.92 (0.35) 1.01 (0.33)** n.m. 21.5 n.m. 1150 8 Control 1999 511 (49) 513 (64) 0.86 (0.27) 0.86 (0.26) 36.8 1850 Thinned 1999 590 (89) 560 (73)** 0.61 (0.25) 0.72 (0.38)** 36.3 23.0 2700 1350 9 Control 2000 565 (77) 570 (62) 0.75 (0.24) 0.74 (0.28) 31.1 2275 Thinned 2000 540 (48) 544 (47) 0.84 (0.34) 0.79 (0.36) 26.0 16.6 2150 1250 Standard deviations are given in parentheses. Average mean ring density and ring width before and after the thinning were compared using t-test. All trees were harvested in 2010. n.m. = data not measured. *P-value < 0.05; **P-value < 0.01; ***P-value < 0.001. No asterisk indicated non-significant difference. Although properties were measured in the control plots both before and after thinning, control plots were not thinned. Table 2 Tree and stand characteristics 10 years before and after the thinning. Region Site number Treatment Year of thinning Mean ring density (kg m−3) Ring width (mm) Basal area (m2 ha−1) Stand density (stems ha−1) Before thinning After thinning Before thinning After thinning Before thinning After thinning Before thinning After thinning North-Shore 1 Control 1997 526 (75) 510 (78) 0.65 (0.20) 0.71 (0.21)* 33.7 2150 Thinned 1997 519 (62) 521 (68) 0.76 (0.32) 0.71 (0.39) 42.6 26.7 2450 1600 2 Control 1997 561 (60) 551 (73) 0.91 (0.38) 0.84 (0.42) 31.8 2975 Thinned 1997 527 (59) 509 (57)** 0.76 (0.31) 0.71 (0.35) 26.8 19.1 2675 1800 3 Control 1997 568 (71) 554 (61) 0.57 (0.21) 0.57 (0.29) 35.9 3425 Thinned 1997 582 (116) 576 (115) 0.81 (0.34) 0.80 (0.36) 38.9 33.0 2600 2050 Lake-Saint-John 4 Control 1998 504 (84) 513 (79) 0.52 (0.28) 0.62 (0.40)* 37.1 2925 Thinned 1998 506 (41) 512 (71) 1.14 (0.47) 1.01 (0.40)* 40.1 29.7 2500 1575 5 Control 1998 524 (41) 554 (44)*** 0.89 (0.41) 0.73 (0.38)*** 36.8 2550 Thinned 1998 513 (46) 513 (41) 1.08 (0.31) 1.07 (0.32) 43.4 24.3 2050 925 6 Control 1998 489 (48) 508 (55)*** 1.41 (0.49) 1.14 (0.46)*** 34.6 1675 Thinned 1998 536 (59) 512 (58)*** 0.79 (0.24) 0.97 (0.41)*** 29.0 25.0 1775 1400 Abitibi 7 Control 1999 570 (68) 587 (72) 1.02 (0.42) 0.85 (0.32)** 26.8 1175 Thinned 1999 529 (47) 539 (53)* 0.92 (0.35) 1.01 (0.33)** n.m. 21.5 n.m. 1150 8 Control 1999 511 (49) 513 (64) 0.86 (0.27) 0.86 (0.26) 36.8 1850 Thinned 1999 590 (89) 560 (73)** 0.61 (0.25) 0.72 (0.38)** 36.3 23.0 2700 1350 9 Control 2000 565 (77) 570 (62) 0.75 (0.24) 0.74 (0.28) 31.1 2275 Thinned 2000 540 (48) 544 (47) 0.84 (0.34) 0.79 (0.36) 26.0 16.6 2150 1250 Region Site number Treatment Year of thinning Mean ring density (kg m−3) Ring width (mm) Basal area (m2 ha−1) Stand density (stems ha−1) Before thinning After thinning Before thinning After thinning Before thinning After thinning Before thinning After thinning North-Shore 1 Control 1997 526 (75) 510 (78) 0.65 (0.20) 0.71 (0.21)* 33.7 2150 Thinned 1997 519 (62) 521 (68) 0.76 (0.32) 0.71 (0.39) 42.6 26.7 2450 1600 2 Control 1997 561 (60) 551 (73) 0.91 (0.38) 0.84 (0.42) 31.8 2975 Thinned 1997 527 (59) 509 (57)** 0.76 (0.31) 0.71 (0.35) 26.8 19.1 2675 1800 3 Control 1997 568 (71) 554 (61) 0.57 (0.21) 0.57 (0.29) 35.9 3425 Thinned 1997 582 (116) 576 (115) 0.81 (0.34) 0.80 (0.36) 38.9 33.0 2600 2050 Lake-Saint-John 4 Control 1998 504 (84) 513 (79) 0.52 (0.28) 0.62 (0.40)* 37.1 2925 Thinned 1998 506 (41) 512 (71) 1.14 (0.47) 1.01 (0.40)* 40.1 29.7 2500 1575 5 Control 1998 524 (41) 554 (44)*** 0.89 (0.41) 0.73 (0.38)*** 36.8 2550 Thinned 1998 513 (46) 513 (41) 1.08 (0.31) 1.07 (0.32) 43.4 24.3 2050 925 6 Control 1998 489 (48) 508 (55)*** 1.41 (0.49) 1.14 (0.46)*** 34.6 1675 Thinned 1998 536 (59) 512 (58)*** 0.79 (0.24) 0.97 (0.41)*** 29.0 25.0 1775 1400 Abitibi 7 Control 1999 570 (68) 587 (72) 1.02 (0.42) 0.85 (0.32)** 26.8 1175 Thinned 1999 529 (47) 539 (53)* 0.92 (0.35) 1.01 (0.33)** n.m. 21.5 n.m. 1150 8 Control 1999 511 (49) 513 (64) 0.86 (0.27) 0.86 (0.26) 36.8 1850 Thinned 1999 590 (89) 560 (73)** 0.61 (0.25) 0.72 (0.38)** 36.3 23.0 2700 1350 9 Control 2000 565 (77) 570 (62) 0.75 (0.24) 0.74 (0.28) 31.1 2275 Thinned 2000 540 (48) 544 (47) 0.84 (0.34) 0.79 (0.36) 26.0 16.6 2150 1250 Standard deviations are given in parentheses. Average mean ring density and ring width before and after the thinning were compared using t-test. All trees were harvested in 2010. n.m. = data not measured. *P-value < 0.05; **P-value < 0.01; ***P-value < 0.001. No asterisk indicated non-significant difference. Although properties were measured in the control plots both before and after thinning, control plots were not thinned. Wood density measurements During the summer of 2010, each tree was felled and 2-cm-thick disks were sampled from the stump (30 cm) and every 105 cm above this to the top of the tree. In the laboratory, one strip along the North radius was sawn from each disk (longitudinal dimension = 1.6 mm and tangential dimension = 10 mm) for wood density measurements. After removing strips from 29 discs affected by rot, a total of 268 radii were measured. The samples were stored in a conditioning chamber until an equilibrium moisture content of 12 per cent was reached. Due to the low proportion of extractives in black spruce (Lohrasebi et al., 1999), no preliminary chemical treatment was performed on the wood samples. Wood density was obtained by X-ray densitometry at a resolution of 40 μm using a QTRS-01X Tree Ring Analyzer (Quintek Measurement Systems Inc., Knoxville, TN, USA). The measured X-ray density was then calibrated using the wood density of the sample obtained from its oven-dry mass and volume at a moisture content of 12 per cent. The transition point between two consecutive rings was established using the maximum change in density of each intra-ring density variation (Genet et al., 2012). The obtained ring width series were then cross-dated at the intra-tree and intra-stand levels using the functions ‘corr.rwl.seg’ and ‘ccf.series.rwl’ from the ‘dplR’ package (Bunn, 2008) of the R statistical programming environment (R Core Team, 2014). We excluded rings with a cambial age (i.e. ring number counted from the pith) lower than 10 (Xiang et al., 2014b). This corresponded to high-density juvenile wood (Telewski, 1989) with strong internal determinism, e.g. where ontogenic patterns are more important than environmental constraints (Meinzer et al., 2011). Because climatic data were available only from 1961, we also excluded rings formed earlier. In total, 10 384 rings were analysed. A summary of the tree characteristics before and after thinning is given in Table 2. Climatic data Site climatic data was obtained with the BIOSIM software, which performs robust climatic interpolations from a network of meteorological stations (Régnière and Bolstad, 1994; Régnière, 1996). From this process we obtained monthly average temperature and total precipitation from 1961 to 2010. The mean annual temperature varied from 0.1°C (standard deviation = 0.9°C) in the Lake Saint-John region to 1.1°C in the Abitibi region (standard deviation = 1°C), while the annual precipitation sum ranged from 925 mm (standard deviation = 105 mm) in the Abitibi region to 974 mm (standard deviation = 97 mm) in the Lake Saint-John region (Figure 2). In addition, intra-annual variations in temperature ranged from −20°C on average in January to +15°C on average in July (Figure 3). Conversely, monthly precipitation was evenly distributed throughout the year, although summer was the wettest season. Figure 2 View largeDownload slide Ombrothermic diagrams for (a) Lake Saint-John, (b) North-Shore and (c) Abitibi regions. Figure 2 View largeDownload slide Ombrothermic diagrams for (a) Lake Saint-John, (b) North-Shore and (c) Abitibi regions. Figure 3 View largeDownload slide Chronologies of (a) mean annual temperatures and (b) annual precipitation sums for Lake Saint-John (solid), North-Shore (dashed) and Abitibi (dot-dashed) regions. Figure 3 View largeDownload slide Chronologies of (a) mean annual temperatures and (b) annual precipitation sums for Lake Saint-John (solid), North-Shore (dashed) and Abitibi (dot-dashed) regions. To make links between the year of ring formation and climate, we considered 20 climatic variables that consisted of 10 average monthly temperatures and 10 sums of monthly precipitation from January to October. We also considered climatic variables of the year prior to the ring formation (from January to December) to account for carry-over climatic effects, i.e. 12 monthly average temperatures and 12 sums of monthly precipitation. In total, 44 climatic variables were investigated. Model development The first step of our analysis was to build models of MRD as a function of ring width and cambial age. This procedure aimed at removing the effect of growth rate on wood density and the low frequency variations (driven by cambial age) in MRD, which is similar to the dendrochronological standardization. The use of statistical models rather than traditional standardization techniques has proven to be consistent (Bontemps and Esper, 2011) and has already been successfully applied (Franceschini et al., 2012). Linear mixed-effects models were used to match the longitudinal structure of the data, with the site, tree and disk considered as hierarchical levels. Models were fitted using the maximum likelihood method. We used the ‘lme’ function of the ‘nlme’ package (Pinheiro et al., 2007) in the R statistical programming environment. Model selection was based on the analysis of the residuals (normality, bias and homoscedasticity), Akaike’s Information Criterion (AIC) and likelihood ratio tests for nested models (Pinheiro and Bates, 2000). In addition, fit indices, such as pseudo-R2, were computed for the fixed part of the models and for each hierarchical level (Parresol, 1999). Three variables were considered: cambial age, ring width and growth unit. The growth unit was used to account for the variability in MRD along the stem and was calculated as: GU = Tree age – Disc age + 1. The model was built sequentially starting with the fixed part. We then specified the random part of the model and finally verified the variance homogeneity. In a first stage, several mathematical functions, including linear, polynomial and inverse combinations, were tested using cambial age, ring width and growth unit. For this stage, only the intercept was included at the disc, tree and site hierarchical levels. This procedure led to model M0: MRDijkl=m+a1⋅CAijkl+a2⋅CAijkl+a3⋅11+RWijkl+μjkl+μjk+μj+εijkl (1) with MRD being the mean ring density in kg m−3, CA the cambial age in years, RW the ring width in mm, m is the model intercept, a1, a2, and a3 are the parameters of the fixed part, μjkl, μjk and μj, are the parameters associated to the model intercept for, respectively, the site, tree and disk hierarchical levels and εijkl is the residual error. The residuals were defined as the observations minus the predictions of the models (i.e. fixed + random parts). Finally, the i, j, k and l the indices were associated, with the annual ring, the disk, the tree and the site, respectively. This model did not include the growth unit variable as this parameter was not found to bring a significant contribution to the models based on the AIC. Once the fixed part of the model was established, the random part was built by considering all combinations of hierarchical levels (one hierarchical level, combination of two hierarchical levels and all three hierarchical levels). Then, the different hierarchical levels were tested by incorporating random effects on each parameter for each level and the model with the lowest AIC was selected. Last, a first-order autocorrelation function (AR1()) was introduced, which improved the AIC. The resulting model (M1) was MRDijkl=m+a1⋅CAijkl+a2⋅CAijkl+a3⋅11+RWijkl+μjk+α2jk⋅CAijkl+μj+α1j⋅CAijkl⋅CAijkl+α3j⋅11+RWijkl+εijkl (2) with notations being the same as Equation (1) and additionally μjk and α2jk the parameters of the hierarchical level associated with the tree random effects, μj, α1j, α2j and α3j the parameters of the hierarchical level associated to the disk and εijkl the residuals of the models, which integrated a first-order autocorrelation function: εi = p ⋅ εi-1 + e with e ~ N(0, σ). The second step of our analysis consisted of identifying the key climatic variables that can be related to the residuals of the M1 model (εM1), i.e. the difference between observed and predicted MRD. This was performed by constructing a forward stepwise selection of variables by considering linear regressions between εM1 and the 44 considered climatic variables. To compensate for the large number of climatic variables tested, we selected models using the Bayesian information criterion (BIC), which leads to more parsimonious models than the AIC (Pinheiro and Bates, 2000). Once this stepwise regression was achieved, we computed the variance inflation factors (VIF) in order to ensure that we had an acceptable level of multicolinearity between our predictor variables. This was performed using the ‘vif’ function from the ‘car’ package (Fox and Weisberg, 2011) in the R statistical programming environment. When the value of one or more VIF exceeded a threshold of 5, the variable with the highest VIF was excluded. This step was repeated until VIF values were below the retained threshold. The result of the procedure led to a model predicting εM1 as a function of the most influential climatic variables. We hereafter refer to this model as M2. The last step of the analysis was to test whether thinning may have influenced the relationship between MRD and (i) cambial age and ring width on the one hand and (ii) climatic variables on the other. We applied the bootstrapped Gershunov test to detect whether variations of the moving-window correlation resulted from intrinsic variability or from true deviation from such variability (Gershunov et al., 2001; Franceschini et al., 2012). Specifically, we first tested the strength of the relationship between MRD and both ring width and cambial age by comparing the predictions of M1 and MRD observations along the time axis (i.e. years since thnning) for rings formed in thinned and control trees. In addition, we aimed to assess whether thinning could have affected the relationship between MRD and climatic variables. Because the relationship between cambial age and ring width was likely to change with thinning, we did not fit a complete model that would have included cambial age, ring width and climatic variables. Instead, the strength of the relationship between εM1 and climatic variables was assessed by comparing predictions of M2 with the observed values (i.e. εM1) along the time axis for rings formed in thinned and control trees. For both tests, it was expected that the moving-window correlation would be similar for thinned and control trees, but would diverge after thinning. Temporal moving-window Spearman’s rank correlation coefficients between predictions and observations of both models were computed using a moving window of 10 years, with a 1-year lag. In order to take into account the different years of thinning on the sites, we considered the time since thinning rather than the calendar year to perform the moving window correlations. The first period investigated ranged from 39 to 30 years before thinning and the last period ranged from 3 to 12 years after thinning. Specifically, the Gershunov test consists of building a bootstrap distribution for the standard deviation of the moving-window correlation by generating 1000 bootstrap samples of 10 randomly selected years. This provides an overall mean of the correlation and its 90 per cent bilateral confidence intervals. These were obtained from the mean and quantiles 0.05 and 0.95 of the 1000 bootstrapped sample correlations of sample size 10 over the study period. This confidence interval allowed us to detect local instability of the moving-window correlation. Results For most sites in all regions, the average tree growth rate and MRD were not significantly different between control and thinned stands (Table 2). Only a few sites had a significant difference in MRD and RW before and after the thinning. For example, a decrease in RW and an increase in MRD were observed in the control plot of site 6 while the reverse was observed in the thinned plot. However, no generalization of this pattern could be made. Similarly, no general trend in MRD was observed in MRD chronologies after thinning (Figure 4). In the case of RW chronologies, post-treatment increases were generally observed in thinned stands, except in sites 1, 2 and 9 (Figure 5). Chronologies also evidenced that in some sites, RW abruptly decreased in both control and thinned stands between 1975 and 1980. Figure 4 View largeDownload slide Chronologies of mean ring density for control (solid line) and thinned (dashed line) stands for sites on (a)–(c) Lake Saint-John, (d)–(f) North-Shore and (g)–(i) Abitibi regions. The vertical solid lines correspond to the year of thinning. Figure 4 View largeDownload slide Chronologies of mean ring density for control (solid line) and thinned (dashed line) stands for sites on (a)–(c) Lake Saint-John, (d)–(f) North-Shore and (g)–(i) Abitibi regions. The vertical solid lines correspond to the year of thinning. Figure 5 View largeDownload slide Chronologies of ring width for control (solid line) and thinned (dashed line) stands for sites on (a)–(c) Lake Saint- John, (d)–(f) North-Shore and (g)–(i) Abitibi regions. The vertical solid lines correspond to the year of thinning. Figure 5 View largeDownload slide Chronologies of ring width for control (solid line) and thinned (dashed line) stands for sites on (a)–(c) Lake Saint- John, (d)–(f) North-Shore and (g)–(i) Abitibi regions. The vertical solid lines correspond to the year of thinning. Data observation (Figure 6) and the results of the model fitting process (Table 3) showed that cambial age and ring width had, respectively, a positive and a negative effect on MRD. The mathematical form of the effect of cambial age reflects the fact that MRD seemed to stabilize after reaching an age of ~60 years. A large part of the total variation was attributable to inter- and intra-stem random variation, as demonstrated by the large standard deviations of the random effects parameters at the tree and disc levels. The residual standard error of the model was 36.4 kg m−3. The pseudo-R2 of model M1 showed that 14.9 per cent of the variation in MRD was explained by the fixed part only, but this increased to 77.4 per cent when taking into account the tree and disk random effects. Table 3 Summary statistics of model M1. Variables Parameter estimates Standard deviation Degrees of freedom Fixed Intercept 383 17.0 10 039 CA 0.237 0.518 10 039 CA 5.14 5.91 10 039 11+RW 196 23.9 10 039 Random effect associated to the tree Intercept 37.3 CA 7.76 Random effect associated to the disk Intercept 162 CA 5.34 CA 59.2 11+RW 325 Autocorrelation 0.393 Residuals 36.4 AIC 101434 BIC 101 572 Pseudo-R2 (%) Fixed 14.9 Random tree 30.8 Random disk 31.7 Total 77.4 Variables Parameter estimates Standard deviation Degrees of freedom Fixed Intercept 383 17.0 10 039 CA 0.237 0.518 10 039 CA 5.14 5.91 10 039 11+RW 196 23.9 10 039 Random effect associated to the tree Intercept 37.3 CA 7.76 Random effect associated to the disk Intercept 162 CA 5.34 CA 59.2 11+RW 325 Autocorrelation 0.393 Residuals 36.4 AIC 101434 BIC 101 572 Pseudo-R2 (%) Fixed 14.9 Random tree 30.8 Random disk 31.7 Total 77.4 Table 3 Summary statistics of model M1. Variables Parameter estimates Standard deviation Degrees of freedom Fixed Intercept 383 17.0 10 039 CA 0.237 0.518 10 039 CA 5.14 5.91 10 039 11+RW 196 23.9 10 039 Random effect associated to the tree Intercept 37.3 CA 7.76 Random effect associated to the disk Intercept 162 CA 5.34 CA 59.2 11+RW 325 Autocorrelation 0.393 Residuals 36.4 AIC 101434 BIC 101 572 Pseudo-R2 (%) Fixed 14.9 Random tree 30.8 Random disk 31.7 Total 77.4 Variables Parameter estimates Standard deviation Degrees of freedom Fixed Intercept 383 17.0 10 039 CA 0.237 0.518 10 039 CA 5.14 5.91 10 039 11+RW 196 23.9 10 039 Random effect associated to the tree Intercept 37.3 CA 7.76 Random effect associated to the disk Intercept 162 CA 5.34 CA 59.2 11+RW 325 Autocorrelation 0.393 Residuals 36.4 AIC 101434 BIC 101 572 Pseudo-R2 (%) Fixed 14.9 Random tree 30.8 Random disk 31.7 Total 77.4 Figure 6 View largeDownload slide Mean ring density as a function of (a) cambial age and (b) ring width. The solid lines correspond to a loess smoother. Figure 6 View largeDownload slide Mean ring density as a function of (a) cambial age and (b) ring width. The solid lines correspond to a loess smoother. When performing the forward stepwise regressions, the residuals of model M1, εM1, were found to be related to several climatic variables (Figure 7). When εM1 was related to all the climatic variables (model M2), the residual standard error of the model was 32.7 kg m−3 and the model accounted for only 8.0 per cent of the variation. The VIFs were checked and all ranged between 1 and 2 (not shown), leading to the conclusion that the level of multicollinearity in the model was acceptable. Overall, εM1 increased with decreasing temperatures of the fall and winter before the year of ring formation (previous October, January and February). In addition, spring temperatures (in April and May) were positively related to εM1, while summer temperatures (July and August) were negatively related to εM1. Other temperature variables were found to be significantly related to εM1 but no general trend could be inferred: June and August temperatures of the previous growing season were positively related to εM1, while it was the opposite for July temperatures of the previous growing season. When considering the effects of precipitation on εM1, results showed that monthly precipitation from June to August were positively related to εM1. The other effects of precipitation were unclear as there were no consecutive months for which significant effects had the same sign. Figure 7 View largeDownload slide Parameter values from the forward stepwise regression model of the residuals from M1 for (a) temperatures and (b) precipitation variables from January of the year previous to ring formation until October of the year of ring formation. All parameters had a P-value lower than 10−3, except August temperatures which had a P-value between 10−2 and 10−3. The letter p before the names of months refers to monthly climatic variables of the year previous to ring formation. The vertical dotted line separates climatic variablee of the year previous to ring formation and the year of ring formation. Figure 7 View largeDownload slide Parameter values from the forward stepwise regression model of the residuals from M1 for (a) temperatures and (b) precipitation variables from January of the year previous to ring formation until October of the year of ring formation. All parameters had a P-value lower than 10−3, except August temperatures which had a P-value between 10−2 and 10−3. The letter p before the names of months refers to monthly climatic variables of the year previous to ring formation. The vertical dotted line separates climatic variablee of the year previous to ring formation and the year of ring formation. The moving window Spearman’s rank correlation coefficient from model M1 showed that the correlation between observed MRD and predictions of model M1 was higher in the control stands than in the thinned stands for almost all the observation period (Figure 8a). The variations of the correlation coefficient were synchronous before thinning between thinned and control stands, but then diverged after thinning. Specifically, the correlation coefficient between MRD observations and predictions of the M1 model increased from 0.891 2 years before thinning to 0.923 3 years after thinning in control stands. Because the Gershunov test indicated that the average correlation coefficient was 0.897 with a confidence interval ranging from 0.883 to 0.911, this increase in the correlation coefficient significantly remained out its intrinsic range of variability for 5 consecutive years. Six years after thinning, the correlation coefficient decreased to within this intrinsic range of variability. In thinned stands, the correlation coefficient decreased from 0.879 2 years before thinning to a minimum of 0.847 6 years after thinning. In these stands, the Gershunov test indicated that the average correlation coefficient was 0.867 with a confidence interval ranging from 0.851 to 0.882. In this case, the correlation coefficient was lower than its intrinsic range of variability during a 4-year period, i.e. from 4 years after thinning to 7 years after thinning. Afterwards, the correlation coefficient increased to within the intrinsic range of variability. Figure 8 View largeDownload slide Moving-window Spearman’s rank correlation coefficients between (a) mean ring density observations and predictions of the M1 model and (b) residuals of the M1 model and predictions of M2. Solid and dashed lines correspond, respectively, to the local correlation computed for control and thinned stands on a moving window of 10 years with a 1-year lag. Each period is centred so that year i corresponds to the period (i − 4; i + 5). The light grey and dark grey envelopes correspond to the 90% bilateral confidence intervals of the moving-window correlations for control and thinned stands, respectively (Gershunov test, see section Materials and methods). Figure 8 View largeDownload slide Moving-window Spearman’s rank correlation coefficients between (a) mean ring density observations and predictions of the M1 model and (b) residuals of the M1 model and predictions of M2. Solid and dashed lines correspond, respectively, to the local correlation computed for control and thinned stands on a moving window of 10 years with a 1-year lag. Each period is centred so that year i corresponds to the period (i − 4; i + 5). The light grey and dark grey envelopes correspond to the 90% bilateral confidence intervals of the moving-window correlations for control and thinned stands, respectively (Gershunov test, see section Materials and methods). The moving window Spearman’s rank correlation coefficient between εM1 and residuals of model M2 tended to be higher in thinned stands than control stands for most of the time period considered (Figure 8b). Except for short periods (e.g. from 3 years before thinning until the year of thinning), the variations of the correlation coefficient were synchronous in both control and thinned stands. In control stands, the average moving-window correlation coefficient between εM1 and residuals of model M2 was 0.241, with a confidence interval ranging from 0.150 to 0.326. Except for the period extending from 33 to 30 years before thinning, the correlation coefficient remained within its intrinsic range of variability, even after thinning. For thinned stands, the average moving-window correlation coefficient was 0.309 with an intrinsic range of variability spanning from 0.231 to 0.378. The correlation coefficient was below the lower bound from 1 year before thinning to 3 years after thinning. This is concomitant with the short loss of synchronicity in the variations of the correlation coefficients between control and thinned stands. Discussion Effects of climatic variables on MRD Overall, temperatures of the previous winter, as well as precipitation of the previous year, had negative effects on εM1. This illustrates the importance of carry-over effects on wood formation (Barbaroux and Bréda, 2002; Kagawa et al., 2006), as already reported for trees growing at high altitude (e.g. Picea crassifolia Kom., Xu et al., 2012), boreal (e.g. P. mariana (Mill.) B.S.P., Xiang et al., 2014a), and temperate conditions (e.g. Picea Rubens Sarg., Conkey, 1979). Although the effect of temperatures of the previous winter on εM1 may appear curious, this may also be linked to a carry-over process (Barbaroux and Bréda, 2002). In the case of high winter precipitation, the following snowmelt in spring may increase water availability, which favours the production of cells with large lumen and may thus cause a decrease in εM1 (Fritts et al., 1991). The positive effect of spring temperatures on the mean ring density of black spruce trees was previously observed by Xiang et al. (2014a) and can be related to cambial phenology. Indeed, black spruce cambial activity is under strong influence of spring climatic conditions (Rossi et al., 2011), and is stimulated by favourable temperatures during the beginning of the growing season (Wang et al., 2002). Such observations have been generalized to all conifers growing in northern climatic conditions (Deslauriers and Morin, 2005; Rossi et al., 2008). These studies suggest that high spring temperatures may hasten the onset of cambial activity and stimulate both cell division and cell wall production in earlywood, therefore leading to higher earlywood and mean ring densities. The effect of current summer temperatures and precipitation is in accordance with the results of Xiang et al. (2014a). First, warm summers are often associated with high potential evapotranspiration that can reduce tree photosynthesis (Lebourgeois et al., 1998; Flexas et al., 2006), shorten the period of wood formation (Sohn et al., 2012) and ultimately decrease carbohydrate allocation to cell wall production. This would result in lower wood density, as evidenced in the present study. Second, we showed that summer precipitation was negatively associated with mean ring density. In the present study, precipitation was more abundant during the summer months, which likely increased water availability. Such high water availability is known to result in the production of cells with larger lumens (Fritts et al., 1991) as cell enlargement is still ongoing during latewood formation (Deslauriers et al., 2003). This ultimately leads to rings with lower wood density. Effects of thinning on MRD In this study, only a small effect of thinning on mean ring density was detected, while its effect on growth rate was generally positive. For numerous species, wood density decreases following thinning as a consequence of a higher growth rate (Pinus banksiana Lamb., Barbour et al., 1994; Schneider et al., 2008; Pinus sylvestris L., Peltola et al., 2007). However, in line with our results, the wood density response of black spruce is generally limited after partial cutting treatments such as careful logging around small merchantable stems (Lemay et al., 2016) and commercial thinning (Vincent et al., 2011). This can explain the decline in the strength of the relationship between mean ring density and growth rate that was evidenced by the correlation coefficient between predictions of model M1 and mean ring density observations. As inter-tree competition decreases following thinning, residual trees have better access to growth resources. In turn, thinned trees could use these resources to invest in secondary growth, which is a lower carbon allocation priority than height growth (Mccarthy and Enquist, 2007). This is consistent with our observation that the strength of the relationship between mean ring density and growth rate decreased after thinning. This potentially higher secondary growth production of thinned trees may be used to produce rings with higher proportion of earlywood (Wang et al., 2002; St-Germain and Krause, 2008) but also higher earlywood density, which would lead to a limited change in mean ring density. We were unfortunately unable to test this hypothesis because the slow growth rate of the sampled trees prevented us from differentiating the earlywood from the latewood (14 per cent of the sampled rings were narrower than 0.5 mm). It may also be that at such slow growth rates, an increase in growth rate leads to a similar increase of both the earlywood and latewood widths. Effect of thinning on the relationship between MRD and climatic variables The present study showed similar relationships between mean ring density and climatic variables for both control and thinned plots. MRD was under the influence of several climatic variables, similarly to the few previous studies conducted in temperate or boreal conditions (Wimmer and Grabner, 2000; Xiang et al., 2014b). Although these effects could be detected, they represented only a small proportion of the variation explained by the M2 model, which may explain why thinning did not influence the relationship between MRD and climatic variables. Such low influence of climatic variables on MRD may result from the presence of compression wood in discs. Specifically, compression wood may lead to unexplained variation in mean ring density that would be included in the residuals of the models. In addition, the small proportion of the MRD variation explained by climatic variables may result from the fact that we considered mean ring density. Latewood density is under a stronger climatic influence than earlywood density (Wimmer and Grabner, 2000) and both ring components respond differently to climatic variables (Franceschini et al., 2013b). Nonetheless, the limited dependence of MRD on climatic variables could also imply that some of the effects highlighted in the present study may have been detected, but they did not necessarily imply a causal relationship (Shipley, 2016). For example, the temperatures observed during June, July and August of the previous year were, respectively, associated with positive, negative and positive parameters in model M2 (Figure 7a). In addition, the climatic dependence of black spruce mean ring density in boreal conditions may not be strong enough to be modulated by thinning. Our sample trees were growing in conditions where climate has a lower influence on wood density than at the treeline (Gort-Oromi et al., 2011; Xiang et al., 2014a). While we observed rather large intra- and inter-annual variation in temperature, the growth conditions may not have been sufficiently limiting to affect MRD significantly. Intra-annual variations in precipitation showed that drought events are unlikely, and thus it is understandable that this climatic variable does not strongly influence the mean ring density. It is widely recognized that the annual ring characteristics of trees growing in harsh environments are strongly dependent on climatic variables. For example, summer temperatures are critical at the latitudinal treeline (Wang et al., 2002; Pritzkow et al., 2014), whereas tree ring indicators are influenced by summer droughts in the Mediterranean context (Büntgen et al., 2010; Rozas et al., 2011). Such strong climatic dependence of tree characteristics may reflect the fact that the growth limitation in harsh environments is mainly caused by the meristem activity (sink limitation) while trees growing away from the treeline are more limited by substrate availability (source limitation; Körner, 1998; Salzer et al., 2009). In addition, black spruce is a widespread species in North-America with a large ecological range (Burns and Honkala, 1990), which could reflect a lower sensitivity to external factors. Conclusion We found that mean ring density was unaffected by a slight increase in ring width following thinning in black spruce stands. This produced a decrease in the strength of the relationship between mean ring density and annual ring width. This was interpreted as an effect of better resource repartition in the stand. Further work is needed to consider temporal variations of photosynthetic capacity as a consequence of growth development and its effects on wood density. However, our second hypothesis was not confirmed, as no change in the effects of climatic variables on mean ring density was found. It is possible that (i) trees were not limited enough by climatic variables to exhibit such change and (ii) the large black spruce ecological niche makes the climatic dependence of MRD unaffected by thinning. Further testing of this hypothesis should be done both in temperate and in harsher climatic conditions. From a management point of view, the maintenance of black spruce wood density with increasing growth rate could lead to an increase in wood value after thinning. Although this would be attractive to both forest managers and wood processors, a full assessment of the economical returns of commercial thinning in these stands remains to be made. Acknowledgements The authors would like to thank Stephane Tremblay for giving access to the thinning trial data, Mikael Bernier and Marie-Pier Arsenault for field measurements, Mélanie Desrochers (CEF) for producing the site location map, Laurence Martel for wood density measurements, and the two anonymous reviewers and the associated editor for their helpful comments and suggestions on an earlier version of the manuscript. Funding This work was supported by the Fonds de Recherche du Québec – Nature et technologies and the Natural Sciences and Engineering Research Council of Canada Conflict of interest statement None declared. References Aussenac , G. and Granier , A. 1988 Effects of thinning on water stress and growth in Douglas-fir . Can. J. For. Res. 18 , 100 – 105 . doi:10.1139/x88-015 . Google Scholar CrossRef Search ADS Barbaroux , C. and Bréda , N. 2002 Contrasting distribution and seasonal dynamics of carbohydrate reserves in stem wood of adult ring-porous sessile oak and diffuse-porous beech trees . Tree Physiol. 22 , 1201 – 1210 . doi:10.1093/treephys/22.17.1201 . Google Scholar CrossRef Search ADS PubMed Barbour , R.J. , Fayle , D.C.F. , Chauret , G. , Cook , J. , Karsh , M.B. and Ran , S. 1994 Breast-height relative density and radial growth in mature jack pine (Pinus banksiana) for 38 years after thinning . Can. J. For. Res. 24 , 2439 – 2447 . doi:10.1139/x94-315 . Google Scholar CrossRef Search ADS Bontemps , J.-D. and Esper , J. 2011 Statistical modelling and RCS detrending methods provide similar estimates of long-term trend in radial growth of common beech in north-eastern France . Dendrochronologia 29 , 99 – 107 . doi:10.1016/j.dendro.2010.09.002 . Google Scholar CrossRef Search ADS Bouriaud , O. , Bréda , N. , Dupouey , J.-L. and Granier , A. 2005 Is ring width a reliable proxy for stem-biomass increment? A case study in European beech . Can. J. For. Res. 35 , 2920 – 2933 . doi:10.1139/x05-202 . Google Scholar CrossRef Search ADS Brazier , J.D. 1970 Timber improvementII. The effect of vigour on young-growth sitka spruce . For. Int. J. For. Res. 43 , 135 – 150 . doi:10.1093/forestry/43.2.135 . Bunn , A.G. 2008 A dendrochronology program library in R (dplR) . Dendrochronologia 26 , 115 – 124 . doi:10.1016/j.dendro.2008.01.002 . Google Scholar CrossRef Search ADS Büntgen , U. , Frank , D. , Trouet , V. and Esper , J. 2010 Diverse climate sensitivity of Mediterranean tree-ring width and density . Trees 24 , 261 – 273 . doi:10.1007/s00468-009-0396-y . Google Scholar CrossRef Search ADS Burns , R.M. and Honkala , B.H. , Technical C . 1990 . Silvics of North America. Volume 1. Conifers. vi + 675 pp. Bustos , C. , Beauregard , R. , Mohammad , M. and Hernandez , R.E. 2003 Structural performance of finger-jointed black spruce lumber with different joint configurations . For. Prod. J. 53 , 72 – 76 . Conkey , L.E. 1979 Response of tree-ring density to climate in Maine, U.S.A . Tree-Ring Bull. 39 , 29 – 38 . D’Amato , A.W. , Bradford , J.B. , Fraver , S. and Palik , B.J. 2013 Effects of thinning on drought vulnerability and climate response in north temperate forest ecosystems . Ecol. Appl. 23 , 1735 – 1742 . doi:10.1890/13-0677.1 . Google Scholar CrossRef Search ADS PubMed Deslauriers , A. and Morin , H. 2005 Intra-annual tracheid production in balsam fir stems and the effect of meteorological variables . Trees 19 , 402 – 408 . doi:10.1007/s00468-004-0398-8 . Google Scholar CrossRef Search ADS Deslauriers , A. , Morin , H. and Begin , Y. 2003 Cellular phenology of annual ring formation of Abies balsamea in the Quebec boreal forest (Canada) . Can. J. For. Res. 33 , 190 – 200 . doi:10.1139/x02-178 . Google Scholar CrossRef Search ADS Filipescu , C.N. , Lowell , E.C. , Koppenaal , R. and Mitchell , A.K. 2013 Modeling regional and climatic variation of wood density and ring width in intensively managed Douglas-fir . Can. J. For. Res. 44 , 220 – 229 . doi:10.1139/cjfr-2013-0275 . Google Scholar CrossRef Search ADS Flexas , J. , Bota , J. , Galmés , J. , Medrano , H. and Ribas-Carbó , M. 2006 Keeping a positive carbon balance under adverse conditions: responses of photosynthesis and respiration to water stress . Physiol. Plant. 127 , 343 – 352 . doi:10.1111/j.1399-3054.2006.00621.x . Google Scholar CrossRef Search ADS Fox , J. and Weisberg , S. 2011 An {R} Companion to Applied Regression . 2nd edn . Sage . Franceschini , T. , Bontemps , J.-D. and Leban , J.-M. 2012 Transient historical decrease in earlywood and latewood density and unstable sensitivity to summer temperature for Norway spruce in northeastern France . Can. J. For. Res. 42 , 219 – 226 . doi:10.1139/x11-182 . Google Scholar CrossRef Search ADS Franceschini , T. , Bontemps , J.-D. , Perez , V. and Leban , J.-M. 2013 a Divergence in latewood density response of Norway spruce to temperature is not resolved by enlarged sets of climatic predictors and their non-linearities . Agric. For. Meteorol. 180 , 132 – 141 . doi:10.1016/j.agrformet.2013.05.011 . Google Scholar CrossRef Search ADS Franceschini , T. , Longuetaud , F. , Bontemps , J.-D. , Bouriaud , O. , Caritey , B.-D. and Leban , J.-M. 2013 b Effect of ring width, cambial age, and climatic variables on the within-ring wood density profile of Norway spruce Picea abies (L.) Karst . Trees Struct. Funct. 27 , 913 – 925 . doi:10.1007/s00468-013-0844-6 . Google Scholar CrossRef Search ADS Fritts , H.C. , Vaganov , E.A. , Sviderskaya , I.V. and Shashkin , A.V. 1991 Climatic variation and tree-ring structure in conifers: empirical and mechanistic models of tree-ring width, number of cells, cell size, cell-wall thickness and wood density . Clim. Res. 1 , 97 – 116 . Google Scholar CrossRef Search ADS Gauthier , S. and Vaillancourt , M.-A. , 2008 . Aménagement écosystémique en forêt Boréale. PUQ. Gebhardt , T. , Häberle , K.-H. , Matyssek , R. , Schulz , C. and Ammer , C. 2014 The more, the better? Water relations of Norway spruce stands after progressive thinning . Agric. For. Meteorol. 197 , 235 – 243 . doi:10.1016/j.agrformet.2014.05.013 . Google Scholar CrossRef Search ADS Genet , A. , Auty , Da. , Achim , A. , Bernier , M. , Pothier , D. and Cogliastro , A. 2012 Consequences of faster growth for wood density in northern red oak (Quercus rubra Liebl.) . Forestry . doi:10.1093/forestry/cps057 . cps057. Gershunov , A. , Schneider , N. and Barnett , T. 2001 Low-frequency modulation of the ENSO–Indian monsoon rainfall relationship: signal or noise? J. Clim. 14 , 2486 – 2492 . doi:10.1175/1520-0442(2001)014<2486:LFMOTE>2.0.CO;2 . Google Scholar CrossRef Search ADS Gort-Oromi , J. , Mehtätalo , L. , Peltola , H. , Zubizarreta-Gerendiain , A. , Pulkkinen , P. and Venäläinen , A. 2011 Effects of spacing and genetic entry on radial growth and ring density development in Scots pine (Pinus sylvestris L.) . Ann. For. Sci. 68 , 1233 . doi:10.1007/s13595-011-0117-8 . Google Scholar CrossRef Search ADS Goudiaby , V. , Brais , S. , Berninger , F. and Schneider , R. 2012 Vertical patterns in specific volume increment along stems of dominant jack pine (Pinus banksiana) and black spruce (Picea mariana) after thinning . Can. J. For. Res. 42 , 733 – 748 . doi:10.1139/x2012-029 . Google Scholar CrossRef Search ADS Hacke , U.G. and Sperry , J.S. 2001 Functional and ecological xylem anatomy . Perspect. Plant Ecol. Evol. Syst. 4 , 97 – 115 . doi:10.1078/1433-8319-00017 . Google Scholar CrossRef Search ADS Hacke , U.G. , Sperry , J.S. , Pockman , W.T. , Davis , S.D. and McCulloh , K.A. 2001 Trends in wood density and structure are linked to prevention of xylem implosion by negative pressure . Oecologia 126 , 457 – 461 . doi:10.1007/s004420100628 . Google Scholar CrossRef Search ADS PubMed Hale , S.E. 2003 The effect of thinning intensity on the below-canopy light environment in a Sitka spruce plantation . For. Ecol. Manage. 179 , 341 – 349 . doi:10.1016/S0378-1127(02)00540-6 . Google Scholar CrossRef Search ADS Hughes , M.K. 2002 Dendrochronology in climatology – the state of the art . Dendrochronologia 20 , 95 – 116 . doi:10.1078/1125-7865-00011 . Google Scholar CrossRef Search ADS Jaakkola , T. , Mäkinen , H. and Saranpää , P. 2005 Wood density in Norway spruce: changes with thinning intensity and tree age . Can. J. For. Res. 35 , 1767 – 1778 . doi:10.1139/x05-118 . Google Scholar CrossRef Search ADS Kagawa , A. , Sugimoto , A. and Maximov , T.C. 2006 13CO2 pulse-labelling of photoassimilates reveals carbon allocation within and between tree rings . Plant Cell. Environ. 29 , 1571 – 1584 . doi:10.1111/j.1365-3040.2006.01533.x . Google Scholar CrossRef Search ADS PubMed Kantavichai , R. , Briggs , D. and Turnblom , E. 2010 Modeling effects of soil, climate, and silviculture on growth ring specific gravity of Douglas-fir on a drought-prone site in Western Washington . For. Ecol. Manage. 259 , 1085 – 1092 . doi:10.1016/j.foreco.2009.12.017 . Google Scholar CrossRef Search ADS Kennedy , R.W. 1995 Coniferous wood quality in the future: concerns and strategies . Wood Sci. Technol. 29 , 321 – 338 . doi:10.1007/BF00202581 . Google Scholar CrossRef Search ADS Kohler , M. , Sohn , J. , Nägele , G. and Bauhus , J. 2010 Can drought tolerance of Norway spruce (Picea abies (L.) Karst.) be increased through thinning? Eur. J. For. Res. 129 , 1109 – 1118 . doi:10.1007/s10342-010-0397-9 . Google Scholar CrossRef Search ADS Körner , C. 1998 A re-assessment of high elevation treeline positions and their explanation . Oecologia 115 , 445 – 459 . doi:10.1007/s004420050540 . Google Scholar CrossRef Search ADS PubMed Lachenbruch , B. , Moore , J.R. and Evans , R. 2011 Radial variation in wood structure and function in woody plants, and hypotheses for its occurrence. In Size- and Age-Related Changes in Tree Structure and Function, Tree Physiology . Meinzer F.C. , Lachenbruch B. and Dawson T.E. (eds). Springer , pp. 121 – 164 . Google Scholar CrossRef Search ADS Lebourgeois , F. , Lévy , G. , Aussenac , G. , Clerc , B. and Willm , F. 1998 Influence of soil drying on leaf water potential, photosynthesis, stomatal conductance and growth in two black pine varieties . Ann. Sci. For. 55 , 287 – 299 . doi:10.1051/forest:19980302 . Google Scholar CrossRef Search ADS Lemay , A. , Krause , C. , Achim , A. and Bégin , J. 2016 Growth and wood quality of black spruce and balsam fir following careful logging around small merchantable stems (CLASS) in the boreal forest of Quebec, Canada . For. Int. J. For. Res , 1 – 12 . doi:10.1093/forestry/cpw060 . Little , E.L.J. , 1971 . Atlas of United States trees, volume 1, conifers and important hardwoods: U.S. Department of Agriculture Miscellaneous Publication 1146, 9 p., 200 maps. Lohrasebi , H. , Mabee , W.E. and Roy , D.N. 1999 Chemistry and pulping feasibility of compression wood in black spruce . J. Wood Chem. Technol. 19 , 13 – 25 . doi:10.1080/02773819909349597 . Google Scholar CrossRef Search ADS Ma , S. , Concilio , A. , Oakley , B. , North , M. and Chen , J. 2010 Spatial variability in microclimate in a mixed-conifer forest before and after thinning and burning treatments . For. Ecol. Manage. 259 , 904 – 915 . doi:10.1016/j.foreco.2009.11.030 . Google Scholar CrossRef Search ADS Macdonald , E. and Hubert , J. 2002 A review of the effects of silviculture on timber quality of Sitka spruce . Forestry 75 , 107 – 138 . doi:10.1093/forestry/75.2.107 . Google Scholar CrossRef Search ADS Mccarthy , M.C. and Enquist , B.J. 2007 Consistency between an allometric approach and optimal partitioning theory in global patterns of plant biomass allocation . Funct. Ecol. 21 , 713 – 720 . doi:10.1111/j.1365-2435.2007.01276.x . Google Scholar CrossRef Search ADS Meinzer , F.C. , Lachenbruch , B. and Dawson , T.E. 2011 Size- and Age-Related Changes in Tree Structure and Function . Springer . Google Scholar CrossRef Search ADS Ministère des Ressources Naturelles 2015 Le guide sylvicole du Québec - Tome II: Les concepts et l’application de la sylviculture . Publications du Québec . Misson , L. , Nicault , A. and Guiot , J. 2003 Effects of different thinning intensities on drought response in Norway spruce (Picea abies (L.) Karst.) . For. Ecol. Manage. 183 , 47 – 60 . doi:10.1016/S0378-1127(03)00098-7 . Google Scholar CrossRef Search ADS MRNFP 2003 Manuel d’aménagement Forestier . 4e édn . Ministère des Ressources naturelles, de la Faune et des Parcs du Québec, Direction des programmes forestiers . ed. Parresol , B.R. 1999 Assessing tree and stand biomass: a review with examples and critical comparisons . For. Sci. 45 , 573 – 593 . Peltola , H. , Kilpeläinen , A. , Sauvala , K. , Räisänen , T. and Ikonen , V.-P. 2007 Effects of early thinning regime and tree status on the radial growth and wood density of Scots pine . Silva Fenn 41 , 489 – 505 . Google Scholar CrossRef Search ADS Pinheiro , J. , Bates , D. and DebRoy , S. 2007 . Linear and nonlinear mixed effects models. R Package Version 3, 57. Pinheiro , J.C. and Bates , D.M. 2000 Mixed-effects Models in S and S-PLUS . Springer . Google Scholar CrossRef Search ADS Pritzkow , C. , Heinrich , I. , Grudd , H. and Helle , G. 2014 Relationship between wood anatomy, tree-ring widths and wood density of Pinus sylvestris L. and climate at high latitudes in northern Sweden . Dendrochronologia . doi:10.1016/j.dendro.2014.07.003 . R Core Team , 2014 . R: A Language and Environment for Statistical Computing. Régnière , J. 1996 Generalized approach to landscape–wide seasonal forecasting with temperature–driven simulation models . Environ. Entomol. 25 , 869 – 881 . Google Scholar CrossRef Search ADS Régnière , J. and Bolstad , P. 1994 Statistical simulation of daily air temperature patterns eastern North America to forecast seasonal events in insect pest management . Environ. Entomol. 23 , 1368 – 1380 . Google Scholar CrossRef Search ADS Rossi , S. , Deslauriers , A. , Griçar , J. , Seo , J.-W. , Rathgeber , C.B. , Anfodillo , T. , et al. 2008 Critical temperatures for xylogenesis in conifers of cold climates . Glob. Ecol. Biogeogr. 17 , 696 – 707 . doi:10.1111/j.1466-8238.2008.00417.x . Google Scholar CrossRef Search ADS Rossi , S. , Morin , H. , Deslauriers , A. and Plourde , P.-Y. 2011 Predicting xylem phenology in black spruce under climate warming . Glob. Change Biol. 17 , 614 – 625 . doi:10.1111/j.1365-2486.2010.02191.x . Google Scholar CrossRef Search ADS Rozas , V. , García-González , I. and Zas , R. 2011 Climatic control of intra-annual wood density fluctuations of Pinus pinaster in NW Spain . Trees 25 , 443 – 453 . doi:10.1007/s00468-010-0519-5 . Google Scholar CrossRef Search ADS Salzer , M.W. , Hughes , M.K. , Bunn , A.G. and Kipfmueller , K.F. 2009 Recent unprecedented tree-ring growth in bristlecone pine at the highest elevations and possible causes . Proc. Natl. Acad. Sci. 106 , 20348 – 20353 . doi:10.1073/pnas.0903029106 . Google Scholar CrossRef Search ADS Schneider , R. , Zhang , S.Y. , Swift , D.E. , Bégin , J. and Lussier , J.-M. 2008 Predicting selected wood properties of jack pine following commercial thinning . Can. J. For. Res. 38 , 2030 – 2043 . doi:10.1139/X08-038 . Google Scholar CrossRef Search ADS Shipley , B. 2016 Cause and Correlation in Biology: A User’s Guide to Path Analysis, Structural Equations and Causal Inference with R . Cambridge University Press . Sohn , J.A. , Kohler , M. , Gessler , A. and Bauhus , J. 2012 Interactions of thinning and stem height on the drought response of radial stem growth and isotopic composition of Norway spruce (Picea abies) . Tree Physiol . doi:10.1093/treephys/tps077 . tps077. Soucy , M. , Lussier , J.-M. and Lavoie , L. 2012 Long-term effects of thinning on growth and yield of an upland black spruce stand . Can. J. For. Res. 42 , 1669 – 1677 . doi:10.1139/x2012-107 . Google Scholar CrossRef Search ADS St-Germain , J.-L. and Krause , C. 2008 Latitudinal variation in tree-ring and wood cell characteristics of Picea mariana across the continuous boreal forest in Quebec . Can. J. For. Res. 38 , 1397 – 1405 . doi:10.1139/X08-008 . Google Scholar CrossRef Search ADS Telewski , F.W. 1989 Structure and function of flexure wood in Abies fraseri . Tree Physiol. 5 , 113 – 121 . doi:10.1093/treephys/5.1.113 . Google Scholar CrossRef Search ADS PubMed Todaro , L. and Macchioni , N. 2010 Wood properties of young Douglas-fir in Southern Italy: results over a 12-year post-thinning period . Eur. J. For. Res. 130 , 251 – 261 . doi:10.1007/s10342-010-0425-9 . Google Scholar CrossRef Search ADS Tong , Q. , Tanguay , F. and Zhang , S.Y. 2011 Impact of commercial thinning on annual radial growth and wood density . Wood Fiber Sci. 43 , 311 – 325 . Vaganov , E.A. , Anchukaitis , K.J. and Evans , M.N. 2011 How well understood are the processes that create dendroclimatic records? A mechanistic model of the climatic control on conifer tree-ring growth dynamics. In Dendroclimatology, Developments in Paleoenvironmental Research . Hughes M.K. , Swetnam T.W. and Diaz H.F. (eds). Springer , pp. 37 – 75 . Vincent , M. , Krause , C. and Koubaa , A. 2011 Variation in black spruce (Picea mariana (Mill.) BSP) wood quality after thinning . Ann. For. Sci. 68 , 1115 – 1125 . doi:10.1007/s13595-011-0127-6 . Google Scholar CrossRef Search ADS Vincent , M. , Krause , C. and Zhang , S.Y. 2009 Radial growth response of black spruce roots and stems to commercial thinning in the boreal forest . Forestry . doi:10.1093/forestry/cpp025 cpp025. Wang , L. , Payette , S. and Bégin , Y. 2002 Relationships between anatomical and densitometric characteristics of black spruce and summer temperature at tree line in northern Quebec . Can. J. For. Res. 32 , 477 – 486 . doi:10.1139/x01-208 . Google Scholar CrossRef Search ADS Wang , L. , Payette , S. and Bégin , Y. 2001 1300-year tree-ring width and density series based on living, dead and subfossil black spruce at tree-line in Subarctic Quebec, Canada . The Holocene 11 , 333 – 341 . doi:10.1191/095968301674769686 . Google Scholar CrossRef Search ADS Wimmer , R. and Grabner , M. 2000 A comparison of tree-ring features in Picea abies as correlated with climate . IAWA J. 21 , 403 – 416 . doi:10.1163/22941932-90000256 . Google Scholar CrossRef Search ADS Xiang , W. , Auty , D. , Franceschini , T. , Leitch , M. and Achim , A. 2014 a Wood density-climate relationships are mediated by dominance class in black spruce (Picea mariana (Mill.) B.S.P.) . Forests 5 , 1163 – 1184 . doi:10.3390/f5061163 . Google Scholar CrossRef Search ADS Xiang , W. , Leitch , M. , Auty , D. , Duchateau , E. and Achim , A. 2014 b Radial trends in black spruce wood density can show an age- and growth-related decline . Ann. For. Sci. 71 , 603 – 615 . doi:10.1007/s13595-014-0363-7 . Google Scholar CrossRef Search ADS Xu , J. , Lu , J. , Bao , F. , Evans , R. and Downes , G.M. 2012 Climate response of cell characteristics in tree rings of Picea crassifolia . Holzforschung 67 , 217 – 225 . doi:10.1515/hf-2011-0144 . Zobel , B.J. and van Buijtenen , J.P. , 1989 . Wood Variation: Its Causes and Control. xv + 363 pp. © Institute of Chartered Foresters, 2017. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Forestry: An International Journal Of Forest Research Oxford University Press

Effect of thinning on the relationship between mean ring density and climate in black spruce (Picea mariana (Mill.)B.S.P.)

Loading next page...
 
/lp/ou_press/effect-of-thinning-on-the-relationship-between-mean-ring-density-and-raxNSui0Sw
Publisher
Institute of Chartered Foresters
Copyright
© Institute of Chartered Foresters, 2017. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.
ISSN
0015-752X
eISSN
1464-3626
D.O.I.
10.1093/forestry/cpx040
Publisher site
See Article on Publisher Site

Abstract

Abstract Relationships between wood density and climatic variables have generally been developed from unmanaged stands near the treeline. Using black spruce (Picea mariana (Mill.) B.S.P.) samples from the managed boreal zone in Canada, we investigated whether the relationship between mean ring density (MRD) and climatic variables is altered by silvicultural practices. We analysed the MRD of 10 384 growth rings from 72 trees sampled among 18 stands (nine thinned, nine controls) across Quebec, Canada. We constructed a mixed-effects model relating MRD to cambial age and ring width (RW). Model residuals (εM1), i.e. the difference between observed and predicted MRD, were then related to monthly temperature and precipitation of the year of ring formation and the year before. After thinning, RW slightly increased while MRD remained constant, thus lowering the strength of the relationship between MRD and RW. εM1 were positively related to spring temperatures and negatively related to summer temperatures and precipitation. No effect of thinning on the relationship between εM1 and climatic variables was observed. The sample trees grew in less limiting conditions than at the treeline so the reduced strength of the relationship between MRD and growth rate in thinned stands may result from a higher photosynthetic capacity. Such results may have implications in forest management as thinning could increase the value of black spruce wood. Introduction Conifer wood density is influenced by a large number of variables. Among them, some result from internal causes, such as ageing or growth rate (Zobel and Buijtenen, 1989), while others originate from external causes, such as climate or competition (Zobel and Buijtenen, 1989; Fritts et al., 1991). For several conifer species, mean ring density (MRD) decreases abruptly from the pith to the juvenile wood, after which it slowly increases in the mature wood (Lachenbruch et al., 2011; Xiang et al., 2014b). These changes are often associated with differences in functional requirements (Lachenbruch et al., 2011): the juvenile wood needs to be flexible but strong, while mature wood needs to be mechanically stiff and ensure high hydraulic conductivity. The radial growth rate (referred hereafter as ‘ring width’) of the stem can also be negatively correlated with wood density, especially in the Picea genus (Macdonald and Hubert, 2002; Xiang et al., 2014b). This is often related to changes in the ring structure: with a higher growth rate, the proportion of dense latewood tends to decrease, which in turn affects MRD (Brazier, 1970; Schneider et al., 2008; Franceschini et al., 2013b). Silvicultural treatments may also influence MRD in conifers (Zobel and Buijtenen, 1989). As a general rule, thinning results in an increase in the rate of volume growth (Goudiaby et al., 2012; Soucy et al., 2012) and radial growth of individual trees (Vincent et al., 2009), which leads to a decrease in wood density (Jaakkola et al., 2005). However, unexplained deviations from this general trend were observed in pine species (Peltola et al., 2007; Schneider et al., 2008) and in Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco, Todaro and Macchioni, 2010). In addition, few studies have observed a decrease in wood density following thinning in spruce species in boreal conditions and when observed, the decrease was generally moderate (Picea abies (L.) Karst., Jaakkola et al., 2005) or non-significant (Picea mariana (Mill.) B.S.P., Tong et al., 2011; Vincent et al., 2011). The effects of climate on wood density have been widely studied, especially in the context of past climatic reconstruction (Wang et al., 2001; Vaganov et al., 2011). Density-based dendroclimatology generally uses the maximum ring density of conifers growing at the altitudinal or latitudinal treeline (Hughes, 2002), because the dense latewood is formed during the summer when the climatic constraints are the strongest (Wang et al., 2002; Deslauriers et al., 2003). However, studies performed on spruce species growing in less limiting conditions (i.e. far from the treeline) also suggested that both latewood density (Franceschini et al., 2013a) and MRD can be under significant climatic determinism (Wimmer and Grabner, 2000; Xiang et al., 2014a). When detailing these effects of climatic variables on wood density, it can be derived that high summer temperatures are generally associated with high latewood proportions, and thus high MRD (Wimmer and Grabner, 2000; Franceschini et al., 2013a). Previous studies also reported negative relationships between MRD and summer precipitation (Wang et al., 2002; Xiang et al., 2014a). The effects of silviculture and climatic variables on MRD have been the subject of several studies. Some studies aimed to model simultaneously the effects of thinning and climatic variables on wood density in Douglas fir (Kantavichai et al., 2010; Filipescu et al., 2013), but potential interactions were not investigated. After thinning, growth conditions of residual trees are generally characterized by increased light (Hale, 2003; Ma et al., 2010) and water availability (Aussenac and Granier, 1988; Gebhardt et al., 2014). Therefore, thinning could modify the relationship between MRD and climatic variables. For example, thinning is known to increase diameter growth under drought conditions compared to unthinned stands in Norway spruce (P. abies (L.) Karst., Kohler et al., 2010; Misson et al., 2003), especially in young stands (D’Amato et al., 2013). In addition, conifers have developed mechanisms to reduce the risk of drought-induced cavitation (i.e. embolism), which consists mainly of producing cells with small lumens (Hacke et al., 2001), thus leading to the production of high-density wood (Hacke and Sperry, 2001). This study investigates the effect of thinning on the relationship between climatic variables and mean ring density in black spruce (P. mariana (Mill.) B.S.P.). This species is both widely distributed in North America (Burns and Honkala, 1990) and has considerable economic importance (Bustos et al., 2003). In Quebec (Canada), commercial thinning has been suggested for these stands as a means to modify stand structure in the ecosystemic management strategy (Gauthier and Vaillancourt, 2008) or to promote intensive wood production (Soucy et al., 2012; Ministère des Ressources Naturelles, 2015). The objectives of our study were thus twofold: (i) to identify the effect of thinning on MRD and (ii) to determine the effects of thinning on the relationship between wood density and climatic variables. We chose to work on MRD, because it is an important indicator of wood quality for black spruce (Kennedy, 1995; Vincent et al., 2011) and an influential variable in wood biomass computations (Bouriaud et al., 2005). We first hypothesized that the strength of the relationship between MRD and growth rate decreases after thinning. Our second hypothesis was that thinning also reduces the strength of the relationship between MRD and climatic variables. Materials and methods Stand characteristics and sampling design We collected data from three regions of Quebec, i.e. from West to East, Abitibi, Lake Saint-John and North Shore (Figure 1). The distance between the sampling locations of Abitibi and North Shore was almost 1000 km. In each region, three stands were selected from a large provincial network of permanent plots. The thinning trials were established by the Ministry of Forests, Wildlife and Parks of Quebec (MFWPQ) to investigate long-term effects of operational commercial thinning on stand development. A total of nine stands (three per site) were selected according to three criteria: (i) the proportion of black spruce trees had to be higher than 75 per cent of the total basal area, (ii) tree age at the year of the thinning had to be between 50 and 60 years and (iii) sites had to be accessible by road. The thinning treatment was a commercial thinning from below, which targeted the removal of the least vigorous trees from dense even-aged stands. In each stand, 400-m2 circular sample plots had been established by the MFWPQ in the treated area. A control plot was also established within a 1-ha untreated block in the vicinity of the thinned plot. The characteristics of the sample stands are described in Table 1. Table 1 Stand and tree characteristics in 2010. Region Site Treatment Longitude (degrees) Latitude (degrees) Tree age DBH (cm) Tree height (m) North-Shore 1 Control stand −69.4511 49.1501 94 (23) 13.9 (1.6) 12.9 (2.0) Thinned stand −69.4510 49.1507 84 (10) 13.9 (3.4) 18.0 (0.3) 2 Control stand −69.4533 49.1520 85 (2) 12.6 (2.4) 18.3 (1.9) Thinned stand −69.4534 49.1525 80 (6) 17.9 (2.9) 15.3 (2.3) 3 Control stand −69.4327 49.1520 70 (13) 11.0 (1.1) 13.6 (1.1) Thinned stand −69.4332 49.1520 82 (7) 13.7 (2.7) 14.5 (0.5) Lake-Saint-John 4 Control stand −70.4740 49.2435 82 (28) 14.7 (1.7) 10.3 (1.8) Thinned stand −70.4740 49.2441 72 (8) 17.0 (1.9) 12.3 (2.8) 5 Control stand −70.4718 49.2409 79 (5) 17.2 (5.6) 11.0 (2.1) Thinned stand −70.4711 49.2408 78 (6) 22.9 (1.5) 13.7 (3.1) 6 Control stand −72.1824 49.1047 62 (5) 15.0 (2.9) 8.9 (1.4) Thinned stand −72.1824 49.1047 66 (3) 17.9 (0.7) 12.5 (1.7) Abitibi 7 Control stand −77.0705 48.2038 65 (6) 14.4 (2.3) 13.5 (2.7) Thinned stand −77.0716 48.2035 73 (7) 17.2 (1.6) 13.1 (1.5) 8 Control stand −77.0459 48.1848 82 (4) 17.8 (1.3) 13.6 (4.3) Thinned stand −77.0447 48.1853 82 (6) 17.3 (2.4) 17.3 (1.3) 9 Control stand −77.5142 48.4831 91 (5) 14.0 (1.0) 16.1 (2.1) Thinned stand −77.5146 48.4825 88 (9) 14.1 (2.0) 16.8 (1.4) Region Site Treatment Longitude (degrees) Latitude (degrees) Tree age DBH (cm) Tree height (m) North-Shore 1 Control stand −69.4511 49.1501 94 (23) 13.9 (1.6) 12.9 (2.0) Thinned stand −69.4510 49.1507 84 (10) 13.9 (3.4) 18.0 (0.3) 2 Control stand −69.4533 49.1520 85 (2) 12.6 (2.4) 18.3 (1.9) Thinned stand −69.4534 49.1525 80 (6) 17.9 (2.9) 15.3 (2.3) 3 Control stand −69.4327 49.1520 70 (13) 11.0 (1.1) 13.6 (1.1) Thinned stand −69.4332 49.1520 82 (7) 13.7 (2.7) 14.5 (0.5) Lake-Saint-John 4 Control stand −70.4740 49.2435 82 (28) 14.7 (1.7) 10.3 (1.8) Thinned stand −70.4740 49.2441 72 (8) 17.0 (1.9) 12.3 (2.8) 5 Control stand −70.4718 49.2409 79 (5) 17.2 (5.6) 11.0 (2.1) Thinned stand −70.4711 49.2408 78 (6) 22.9 (1.5) 13.7 (3.1) 6 Control stand −72.1824 49.1047 62 (5) 15.0 (2.9) 8.9 (1.4) Thinned stand −72.1824 49.1047 66 (3) 17.9 (0.7) 12.5 (1.7) Abitibi 7 Control stand −77.0705 48.2038 65 (6) 14.4 (2.3) 13.5 (2.7) Thinned stand −77.0716 48.2035 73 (7) 17.2 (1.6) 13.1 (1.5) 8 Control stand −77.0459 48.1848 82 (4) 17.8 (1.3) 13.6 (4.3) Thinned stand −77.0447 48.1853 82 (6) 17.3 (2.4) 17.3 (1.3) 9 Control stand −77.5142 48.4831 91 (5) 14.0 (1.0) 16.1 (2.1) Thinned stand −77.5146 48.4825 88 (9) 14.1 (2.0) 16.8 (1.4) Standard deviations are given in parentheses. Table 1 Stand and tree characteristics in 2010. Region Site Treatment Longitude (degrees) Latitude (degrees) Tree age DBH (cm) Tree height (m) North-Shore 1 Control stand −69.4511 49.1501 94 (23) 13.9 (1.6) 12.9 (2.0) Thinned stand −69.4510 49.1507 84 (10) 13.9 (3.4) 18.0 (0.3) 2 Control stand −69.4533 49.1520 85 (2) 12.6 (2.4) 18.3 (1.9) Thinned stand −69.4534 49.1525 80 (6) 17.9 (2.9) 15.3 (2.3) 3 Control stand −69.4327 49.1520 70 (13) 11.0 (1.1) 13.6 (1.1) Thinned stand −69.4332 49.1520 82 (7) 13.7 (2.7) 14.5 (0.5) Lake-Saint-John 4 Control stand −70.4740 49.2435 82 (28) 14.7 (1.7) 10.3 (1.8) Thinned stand −70.4740 49.2441 72 (8) 17.0 (1.9) 12.3 (2.8) 5 Control stand −70.4718 49.2409 79 (5) 17.2 (5.6) 11.0 (2.1) Thinned stand −70.4711 49.2408 78 (6) 22.9 (1.5) 13.7 (3.1) 6 Control stand −72.1824 49.1047 62 (5) 15.0 (2.9) 8.9 (1.4) Thinned stand −72.1824 49.1047 66 (3) 17.9 (0.7) 12.5 (1.7) Abitibi 7 Control stand −77.0705 48.2038 65 (6) 14.4 (2.3) 13.5 (2.7) Thinned stand −77.0716 48.2035 73 (7) 17.2 (1.6) 13.1 (1.5) 8 Control stand −77.0459 48.1848 82 (4) 17.8 (1.3) 13.6 (4.3) Thinned stand −77.0447 48.1853 82 (6) 17.3 (2.4) 17.3 (1.3) 9 Control stand −77.5142 48.4831 91 (5) 14.0 (1.0) 16.1 (2.1) Thinned stand −77.5146 48.4825 88 (9) 14.1 (2.0) 16.8 (1.4) Region Site Treatment Longitude (degrees) Latitude (degrees) Tree age DBH (cm) Tree height (m) North-Shore 1 Control stand −69.4511 49.1501 94 (23) 13.9 (1.6) 12.9 (2.0) Thinned stand −69.4510 49.1507 84 (10) 13.9 (3.4) 18.0 (0.3) 2 Control stand −69.4533 49.1520 85 (2) 12.6 (2.4) 18.3 (1.9) Thinned stand −69.4534 49.1525 80 (6) 17.9 (2.9) 15.3 (2.3) 3 Control stand −69.4327 49.1520 70 (13) 11.0 (1.1) 13.6 (1.1) Thinned stand −69.4332 49.1520 82 (7) 13.7 (2.7) 14.5 (0.5) Lake-Saint-John 4 Control stand −70.4740 49.2435 82 (28) 14.7 (1.7) 10.3 (1.8) Thinned stand −70.4740 49.2441 72 (8) 17.0 (1.9) 12.3 (2.8) 5 Control stand −70.4718 49.2409 79 (5) 17.2 (5.6) 11.0 (2.1) Thinned stand −70.4711 49.2408 78 (6) 22.9 (1.5) 13.7 (3.1) 6 Control stand −72.1824 49.1047 62 (5) 15.0 (2.9) 8.9 (1.4) Thinned stand −72.1824 49.1047 66 (3) 17.9 (0.7) 12.5 (1.7) Abitibi 7 Control stand −77.0705 48.2038 65 (6) 14.4 (2.3) 13.5 (2.7) Thinned stand −77.0716 48.2035 73 (7) 17.2 (1.6) 13.1 (1.5) 8 Control stand −77.0459 48.1848 82 (4) 17.8 (1.3) 13.6 (4.3) Thinned stand −77.0447 48.1853 82 (6) 17.3 (2.4) 17.3 (1.3) 9 Control stand −77.5142 48.4831 91 (5) 14.0 (1.0) 16.1 (2.1) Thinned stand −77.5146 48.4825 88 (9) 14.1 (2.0) 16.8 (1.4) Standard deviations are given in parentheses. Figure 1 View largeDownload slide Site location map. The species distribution map originates from Little (1971). Figure 1 View largeDownload slide Site location map. The species distribution map originates from Little (1971). The commercial thinning was planned to be conducted at least 15 years prior to the final harvest, with a reduction of 28–35 per cent of the initial stand basal area (MRNFP, 2003). Given variations in skid trail proportion between plots, the actual thinning intensity at the plot level ranged between 18 and 44 per cent of the initial stand basal area. Dendrometric measurements on permanent plots were made before and after harvesting. The thinning was conducted between 1997 and 2000 depending on the region and site (Table 2), while the tree sampling for the present study was conducted in 2010. In each plot, the centre of a variable-radius sampling plot was established with a factor 2 prism. Starting from a random azimuth and turning clockwise, the first, third, fifth and seventh merchantable trees (DBH > 9 cm) were selected for destructive sampling. Four trees were sampled in each of the 18 plots (3 regions × 3 sites per region × 2 plots per site) for a total of 72 black spruce trees. Table 2 Tree and stand characteristics 10 years before and after the thinning. Region Site number Treatment Year of thinning Mean ring density (kg m−3) Ring width (mm) Basal area (m2 ha−1) Stand density (stems ha−1) Before thinning After thinning Before thinning After thinning Before thinning After thinning Before thinning After thinning North-Shore 1 Control 1997 526 (75) 510 (78) 0.65 (0.20) 0.71 (0.21)* 33.7 2150 Thinned 1997 519 (62) 521 (68) 0.76 (0.32) 0.71 (0.39) 42.6 26.7 2450 1600 2 Control 1997 561 (60) 551 (73) 0.91 (0.38) 0.84 (0.42) 31.8 2975 Thinned 1997 527 (59) 509 (57)** 0.76 (0.31) 0.71 (0.35) 26.8 19.1 2675 1800 3 Control 1997 568 (71) 554 (61) 0.57 (0.21) 0.57 (0.29) 35.9 3425 Thinned 1997 582 (116) 576 (115) 0.81 (0.34) 0.80 (0.36) 38.9 33.0 2600 2050 Lake-Saint-John 4 Control 1998 504 (84) 513 (79) 0.52 (0.28) 0.62 (0.40)* 37.1 2925 Thinned 1998 506 (41) 512 (71) 1.14 (0.47) 1.01 (0.40)* 40.1 29.7 2500 1575 5 Control 1998 524 (41) 554 (44)*** 0.89 (0.41) 0.73 (0.38)*** 36.8 2550 Thinned 1998 513 (46) 513 (41) 1.08 (0.31) 1.07 (0.32) 43.4 24.3 2050 925 6 Control 1998 489 (48) 508 (55)*** 1.41 (0.49) 1.14 (0.46)*** 34.6 1675 Thinned 1998 536 (59) 512 (58)*** 0.79 (0.24) 0.97 (0.41)*** 29.0 25.0 1775 1400 Abitibi 7 Control 1999 570 (68) 587 (72) 1.02 (0.42) 0.85 (0.32)** 26.8 1175 Thinned 1999 529 (47) 539 (53)* 0.92 (0.35) 1.01 (0.33)** n.m. 21.5 n.m. 1150 8 Control 1999 511 (49) 513 (64) 0.86 (0.27) 0.86 (0.26) 36.8 1850 Thinned 1999 590 (89) 560 (73)** 0.61 (0.25) 0.72 (0.38)** 36.3 23.0 2700 1350 9 Control 2000 565 (77) 570 (62) 0.75 (0.24) 0.74 (0.28) 31.1 2275 Thinned 2000 540 (48) 544 (47) 0.84 (0.34) 0.79 (0.36) 26.0 16.6 2150 1250 Region Site number Treatment Year of thinning Mean ring density (kg m−3) Ring width (mm) Basal area (m2 ha−1) Stand density (stems ha−1) Before thinning After thinning Before thinning After thinning Before thinning After thinning Before thinning After thinning North-Shore 1 Control 1997 526 (75) 510 (78) 0.65 (0.20) 0.71 (0.21)* 33.7 2150 Thinned 1997 519 (62) 521 (68) 0.76 (0.32) 0.71 (0.39) 42.6 26.7 2450 1600 2 Control 1997 561 (60) 551 (73) 0.91 (0.38) 0.84 (0.42) 31.8 2975 Thinned 1997 527 (59) 509 (57)** 0.76 (0.31) 0.71 (0.35) 26.8 19.1 2675 1800 3 Control 1997 568 (71) 554 (61) 0.57 (0.21) 0.57 (0.29) 35.9 3425 Thinned 1997 582 (116) 576 (115) 0.81 (0.34) 0.80 (0.36) 38.9 33.0 2600 2050 Lake-Saint-John 4 Control 1998 504 (84) 513 (79) 0.52 (0.28) 0.62 (0.40)* 37.1 2925 Thinned 1998 506 (41) 512 (71) 1.14 (0.47) 1.01 (0.40)* 40.1 29.7 2500 1575 5 Control 1998 524 (41) 554 (44)*** 0.89 (0.41) 0.73 (0.38)*** 36.8 2550 Thinned 1998 513 (46) 513 (41) 1.08 (0.31) 1.07 (0.32) 43.4 24.3 2050 925 6 Control 1998 489 (48) 508 (55)*** 1.41 (0.49) 1.14 (0.46)*** 34.6 1675 Thinned 1998 536 (59) 512 (58)*** 0.79 (0.24) 0.97 (0.41)*** 29.0 25.0 1775 1400 Abitibi 7 Control 1999 570 (68) 587 (72) 1.02 (0.42) 0.85 (0.32)** 26.8 1175 Thinned 1999 529 (47) 539 (53)* 0.92 (0.35) 1.01 (0.33)** n.m. 21.5 n.m. 1150 8 Control 1999 511 (49) 513 (64) 0.86 (0.27) 0.86 (0.26) 36.8 1850 Thinned 1999 590 (89) 560 (73)** 0.61 (0.25) 0.72 (0.38)** 36.3 23.0 2700 1350 9 Control 2000 565 (77) 570 (62) 0.75 (0.24) 0.74 (0.28) 31.1 2275 Thinned 2000 540 (48) 544 (47) 0.84 (0.34) 0.79 (0.36) 26.0 16.6 2150 1250 Standard deviations are given in parentheses. Average mean ring density and ring width before and after the thinning were compared using t-test. All trees were harvested in 2010. n.m. = data not measured. *P-value < 0.05; **P-value < 0.01; ***P-value < 0.001. No asterisk indicated non-significant difference. Although properties were measured in the control plots both before and after thinning, control plots were not thinned. Table 2 Tree and stand characteristics 10 years before and after the thinning. Region Site number Treatment Year of thinning Mean ring density (kg m−3) Ring width (mm) Basal area (m2 ha−1) Stand density (stems ha−1) Before thinning After thinning Before thinning After thinning Before thinning After thinning Before thinning After thinning North-Shore 1 Control 1997 526 (75) 510 (78) 0.65 (0.20) 0.71 (0.21)* 33.7 2150 Thinned 1997 519 (62) 521 (68) 0.76 (0.32) 0.71 (0.39) 42.6 26.7 2450 1600 2 Control 1997 561 (60) 551 (73) 0.91 (0.38) 0.84 (0.42) 31.8 2975 Thinned 1997 527 (59) 509 (57)** 0.76 (0.31) 0.71 (0.35) 26.8 19.1 2675 1800 3 Control 1997 568 (71) 554 (61) 0.57 (0.21) 0.57 (0.29) 35.9 3425 Thinned 1997 582 (116) 576 (115) 0.81 (0.34) 0.80 (0.36) 38.9 33.0 2600 2050 Lake-Saint-John 4 Control 1998 504 (84) 513 (79) 0.52 (0.28) 0.62 (0.40)* 37.1 2925 Thinned 1998 506 (41) 512 (71) 1.14 (0.47) 1.01 (0.40)* 40.1 29.7 2500 1575 5 Control 1998 524 (41) 554 (44)*** 0.89 (0.41) 0.73 (0.38)*** 36.8 2550 Thinned 1998 513 (46) 513 (41) 1.08 (0.31) 1.07 (0.32) 43.4 24.3 2050 925 6 Control 1998 489 (48) 508 (55)*** 1.41 (0.49) 1.14 (0.46)*** 34.6 1675 Thinned 1998 536 (59) 512 (58)*** 0.79 (0.24) 0.97 (0.41)*** 29.0 25.0 1775 1400 Abitibi 7 Control 1999 570 (68) 587 (72) 1.02 (0.42) 0.85 (0.32)** 26.8 1175 Thinned 1999 529 (47) 539 (53)* 0.92 (0.35) 1.01 (0.33)** n.m. 21.5 n.m. 1150 8 Control 1999 511 (49) 513 (64) 0.86 (0.27) 0.86 (0.26) 36.8 1850 Thinned 1999 590 (89) 560 (73)** 0.61 (0.25) 0.72 (0.38)** 36.3 23.0 2700 1350 9 Control 2000 565 (77) 570 (62) 0.75 (0.24) 0.74 (0.28) 31.1 2275 Thinned 2000 540 (48) 544 (47) 0.84 (0.34) 0.79 (0.36) 26.0 16.6 2150 1250 Region Site number Treatment Year of thinning Mean ring density (kg m−3) Ring width (mm) Basal area (m2 ha−1) Stand density (stems ha−1) Before thinning After thinning Before thinning After thinning Before thinning After thinning Before thinning After thinning North-Shore 1 Control 1997 526 (75) 510 (78) 0.65 (0.20) 0.71 (0.21)* 33.7 2150 Thinned 1997 519 (62) 521 (68) 0.76 (0.32) 0.71 (0.39) 42.6 26.7 2450 1600 2 Control 1997 561 (60) 551 (73) 0.91 (0.38) 0.84 (0.42) 31.8 2975 Thinned 1997 527 (59) 509 (57)** 0.76 (0.31) 0.71 (0.35) 26.8 19.1 2675 1800 3 Control 1997 568 (71) 554 (61) 0.57 (0.21) 0.57 (0.29) 35.9 3425 Thinned 1997 582 (116) 576 (115) 0.81 (0.34) 0.80 (0.36) 38.9 33.0 2600 2050 Lake-Saint-John 4 Control 1998 504 (84) 513 (79) 0.52 (0.28) 0.62 (0.40)* 37.1 2925 Thinned 1998 506 (41) 512 (71) 1.14 (0.47) 1.01 (0.40)* 40.1 29.7 2500 1575 5 Control 1998 524 (41) 554 (44)*** 0.89 (0.41) 0.73 (0.38)*** 36.8 2550 Thinned 1998 513 (46) 513 (41) 1.08 (0.31) 1.07 (0.32) 43.4 24.3 2050 925 6 Control 1998 489 (48) 508 (55)*** 1.41 (0.49) 1.14 (0.46)*** 34.6 1675 Thinned 1998 536 (59) 512 (58)*** 0.79 (0.24) 0.97 (0.41)*** 29.0 25.0 1775 1400 Abitibi 7 Control 1999 570 (68) 587 (72) 1.02 (0.42) 0.85 (0.32)** 26.8 1175 Thinned 1999 529 (47) 539 (53)* 0.92 (0.35) 1.01 (0.33)** n.m. 21.5 n.m. 1150 8 Control 1999 511 (49) 513 (64) 0.86 (0.27) 0.86 (0.26) 36.8 1850 Thinned 1999 590 (89) 560 (73)** 0.61 (0.25) 0.72 (0.38)** 36.3 23.0 2700 1350 9 Control 2000 565 (77) 570 (62) 0.75 (0.24) 0.74 (0.28) 31.1 2275 Thinned 2000 540 (48) 544 (47) 0.84 (0.34) 0.79 (0.36) 26.0 16.6 2150 1250 Standard deviations are given in parentheses. Average mean ring density and ring width before and after the thinning were compared using t-test. All trees were harvested in 2010. n.m. = data not measured. *P-value < 0.05; **P-value < 0.01; ***P-value < 0.001. No asterisk indicated non-significant difference. Although properties were measured in the control plots both before and after thinning, control plots were not thinned. Wood density measurements During the summer of 2010, each tree was felled and 2-cm-thick disks were sampled from the stump (30 cm) and every 105 cm above this to the top of the tree. In the laboratory, one strip along the North radius was sawn from each disk (longitudinal dimension = 1.6 mm and tangential dimension = 10 mm) for wood density measurements. After removing strips from 29 discs affected by rot, a total of 268 radii were measured. The samples were stored in a conditioning chamber until an equilibrium moisture content of 12 per cent was reached. Due to the low proportion of extractives in black spruce (Lohrasebi et al., 1999), no preliminary chemical treatment was performed on the wood samples. Wood density was obtained by X-ray densitometry at a resolution of 40 μm using a QTRS-01X Tree Ring Analyzer (Quintek Measurement Systems Inc., Knoxville, TN, USA). The measured X-ray density was then calibrated using the wood density of the sample obtained from its oven-dry mass and volume at a moisture content of 12 per cent. The transition point between two consecutive rings was established using the maximum change in density of each intra-ring density variation (Genet et al., 2012). The obtained ring width series were then cross-dated at the intra-tree and intra-stand levels using the functions ‘corr.rwl.seg’ and ‘ccf.series.rwl’ from the ‘dplR’ package (Bunn, 2008) of the R statistical programming environment (R Core Team, 2014). We excluded rings with a cambial age (i.e. ring number counted from the pith) lower than 10 (Xiang et al., 2014b). This corresponded to high-density juvenile wood (Telewski, 1989) with strong internal determinism, e.g. where ontogenic patterns are more important than environmental constraints (Meinzer et al., 2011). Because climatic data were available only from 1961, we also excluded rings formed earlier. In total, 10 384 rings were analysed. A summary of the tree characteristics before and after thinning is given in Table 2. Climatic data Site climatic data was obtained with the BIOSIM software, which performs robust climatic interpolations from a network of meteorological stations (Régnière and Bolstad, 1994; Régnière, 1996). From this process we obtained monthly average temperature and total precipitation from 1961 to 2010. The mean annual temperature varied from 0.1°C (standard deviation = 0.9°C) in the Lake Saint-John region to 1.1°C in the Abitibi region (standard deviation = 1°C), while the annual precipitation sum ranged from 925 mm (standard deviation = 105 mm) in the Abitibi region to 974 mm (standard deviation = 97 mm) in the Lake Saint-John region (Figure 2). In addition, intra-annual variations in temperature ranged from −20°C on average in January to +15°C on average in July (Figure 3). Conversely, monthly precipitation was evenly distributed throughout the year, although summer was the wettest season. Figure 2 View largeDownload slide Ombrothermic diagrams for (a) Lake Saint-John, (b) North-Shore and (c) Abitibi regions. Figure 2 View largeDownload slide Ombrothermic diagrams for (a) Lake Saint-John, (b) North-Shore and (c) Abitibi regions. Figure 3 View largeDownload slide Chronologies of (a) mean annual temperatures and (b) annual precipitation sums for Lake Saint-John (solid), North-Shore (dashed) and Abitibi (dot-dashed) regions. Figure 3 View largeDownload slide Chronologies of (a) mean annual temperatures and (b) annual precipitation sums for Lake Saint-John (solid), North-Shore (dashed) and Abitibi (dot-dashed) regions. To make links between the year of ring formation and climate, we considered 20 climatic variables that consisted of 10 average monthly temperatures and 10 sums of monthly precipitation from January to October. We also considered climatic variables of the year prior to the ring formation (from January to December) to account for carry-over climatic effects, i.e. 12 monthly average temperatures and 12 sums of monthly precipitation. In total, 44 climatic variables were investigated. Model development The first step of our analysis was to build models of MRD as a function of ring width and cambial age. This procedure aimed at removing the effect of growth rate on wood density and the low frequency variations (driven by cambial age) in MRD, which is similar to the dendrochronological standardization. The use of statistical models rather than traditional standardization techniques has proven to be consistent (Bontemps and Esper, 2011) and has already been successfully applied (Franceschini et al., 2012). Linear mixed-effects models were used to match the longitudinal structure of the data, with the site, tree and disk considered as hierarchical levels. Models were fitted using the maximum likelihood method. We used the ‘lme’ function of the ‘nlme’ package (Pinheiro et al., 2007) in the R statistical programming environment. Model selection was based on the analysis of the residuals (normality, bias and homoscedasticity), Akaike’s Information Criterion (AIC) and likelihood ratio tests for nested models (Pinheiro and Bates, 2000). In addition, fit indices, such as pseudo-R2, were computed for the fixed part of the models and for each hierarchical level (Parresol, 1999). Three variables were considered: cambial age, ring width and growth unit. The growth unit was used to account for the variability in MRD along the stem and was calculated as: GU = Tree age – Disc age + 1. The model was built sequentially starting with the fixed part. We then specified the random part of the model and finally verified the variance homogeneity. In a first stage, several mathematical functions, including linear, polynomial and inverse combinations, were tested using cambial age, ring width and growth unit. For this stage, only the intercept was included at the disc, tree and site hierarchical levels. This procedure led to model M0: MRDijkl=m+a1⋅CAijkl+a2⋅CAijkl+a3⋅11+RWijkl+μjkl+μjk+μj+εijkl (1) with MRD being the mean ring density in kg m−3, CA the cambial age in years, RW the ring width in mm, m is the model intercept, a1, a2, and a3 are the parameters of the fixed part, μjkl, μjk and μj, are the parameters associated to the model intercept for, respectively, the site, tree and disk hierarchical levels and εijkl is the residual error. The residuals were defined as the observations minus the predictions of the models (i.e. fixed + random parts). Finally, the i, j, k and l the indices were associated, with the annual ring, the disk, the tree and the site, respectively. This model did not include the growth unit variable as this parameter was not found to bring a significant contribution to the models based on the AIC. Once the fixed part of the model was established, the random part was built by considering all combinations of hierarchical levels (one hierarchical level, combination of two hierarchical levels and all three hierarchical levels). Then, the different hierarchical levels were tested by incorporating random effects on each parameter for each level and the model with the lowest AIC was selected. Last, a first-order autocorrelation function (AR1()) was introduced, which improved the AIC. The resulting model (M1) was MRDijkl=m+a1⋅CAijkl+a2⋅CAijkl+a3⋅11+RWijkl+μjk+α2jk⋅CAijkl+μj+α1j⋅CAijkl⋅CAijkl+α3j⋅11+RWijkl+εijkl (2) with notations being the same as Equation (1) and additionally μjk and α2jk the parameters of the hierarchical level associated with the tree random effects, μj, α1j, α2j and α3j the parameters of the hierarchical level associated to the disk and εijkl the residuals of the models, which integrated a first-order autocorrelation function: εi = p ⋅ εi-1 + e with e ~ N(0, σ). The second step of our analysis consisted of identifying the key climatic variables that can be related to the residuals of the M1 model (εM1), i.e. the difference between observed and predicted MRD. This was performed by constructing a forward stepwise selection of variables by considering linear regressions between εM1 and the 44 considered climatic variables. To compensate for the large number of climatic variables tested, we selected models using the Bayesian information criterion (BIC), which leads to more parsimonious models than the AIC (Pinheiro and Bates, 2000). Once this stepwise regression was achieved, we computed the variance inflation factors (VIF) in order to ensure that we had an acceptable level of multicolinearity between our predictor variables. This was performed using the ‘vif’ function from the ‘car’ package (Fox and Weisberg, 2011) in the R statistical programming environment. When the value of one or more VIF exceeded a threshold of 5, the variable with the highest VIF was excluded. This step was repeated until VIF values were below the retained threshold. The result of the procedure led to a model predicting εM1 as a function of the most influential climatic variables. We hereafter refer to this model as M2. The last step of the analysis was to test whether thinning may have influenced the relationship between MRD and (i) cambial age and ring width on the one hand and (ii) climatic variables on the other. We applied the bootstrapped Gershunov test to detect whether variations of the moving-window correlation resulted from intrinsic variability or from true deviation from such variability (Gershunov et al., 2001; Franceschini et al., 2012). Specifically, we first tested the strength of the relationship between MRD and both ring width and cambial age by comparing the predictions of M1 and MRD observations along the time axis (i.e. years since thnning) for rings formed in thinned and control trees. In addition, we aimed to assess whether thinning could have affected the relationship between MRD and climatic variables. Because the relationship between cambial age and ring width was likely to change with thinning, we did not fit a complete model that would have included cambial age, ring width and climatic variables. Instead, the strength of the relationship between εM1 and climatic variables was assessed by comparing predictions of M2 with the observed values (i.e. εM1) along the time axis for rings formed in thinned and control trees. For both tests, it was expected that the moving-window correlation would be similar for thinned and control trees, but would diverge after thinning. Temporal moving-window Spearman’s rank correlation coefficients between predictions and observations of both models were computed using a moving window of 10 years, with a 1-year lag. In order to take into account the different years of thinning on the sites, we considered the time since thinning rather than the calendar year to perform the moving window correlations. The first period investigated ranged from 39 to 30 years before thinning and the last period ranged from 3 to 12 years after thinning. Specifically, the Gershunov test consists of building a bootstrap distribution for the standard deviation of the moving-window correlation by generating 1000 bootstrap samples of 10 randomly selected years. This provides an overall mean of the correlation and its 90 per cent bilateral confidence intervals. These were obtained from the mean and quantiles 0.05 and 0.95 of the 1000 bootstrapped sample correlations of sample size 10 over the study period. This confidence interval allowed us to detect local instability of the moving-window correlation. Results For most sites in all regions, the average tree growth rate and MRD were not significantly different between control and thinned stands (Table 2). Only a few sites had a significant difference in MRD and RW before and after the thinning. For example, a decrease in RW and an increase in MRD were observed in the control plot of site 6 while the reverse was observed in the thinned plot. However, no generalization of this pattern could be made. Similarly, no general trend in MRD was observed in MRD chronologies after thinning (Figure 4). In the case of RW chronologies, post-treatment increases were generally observed in thinned stands, except in sites 1, 2 and 9 (Figure 5). Chronologies also evidenced that in some sites, RW abruptly decreased in both control and thinned stands between 1975 and 1980. Figure 4 View largeDownload slide Chronologies of mean ring density for control (solid line) and thinned (dashed line) stands for sites on (a)–(c) Lake Saint-John, (d)–(f) North-Shore and (g)–(i) Abitibi regions. The vertical solid lines correspond to the year of thinning. Figure 4 View largeDownload slide Chronologies of mean ring density for control (solid line) and thinned (dashed line) stands for sites on (a)–(c) Lake Saint-John, (d)–(f) North-Shore and (g)–(i) Abitibi regions. The vertical solid lines correspond to the year of thinning. Figure 5 View largeDownload slide Chronologies of ring width for control (solid line) and thinned (dashed line) stands for sites on (a)–(c) Lake Saint- John, (d)–(f) North-Shore and (g)–(i) Abitibi regions. The vertical solid lines correspond to the year of thinning. Figure 5 View largeDownload slide Chronologies of ring width for control (solid line) and thinned (dashed line) stands for sites on (a)–(c) Lake Saint- John, (d)–(f) North-Shore and (g)–(i) Abitibi regions. The vertical solid lines correspond to the year of thinning. Data observation (Figure 6) and the results of the model fitting process (Table 3) showed that cambial age and ring width had, respectively, a positive and a negative effect on MRD. The mathematical form of the effect of cambial age reflects the fact that MRD seemed to stabilize after reaching an age of ~60 years. A large part of the total variation was attributable to inter- and intra-stem random variation, as demonstrated by the large standard deviations of the random effects parameters at the tree and disc levels. The residual standard error of the model was 36.4 kg m−3. The pseudo-R2 of model M1 showed that 14.9 per cent of the variation in MRD was explained by the fixed part only, but this increased to 77.4 per cent when taking into account the tree and disk random effects. Table 3 Summary statistics of model M1. Variables Parameter estimates Standard deviation Degrees of freedom Fixed Intercept 383 17.0 10 039 CA 0.237 0.518 10 039 CA 5.14 5.91 10 039 11+RW 196 23.9 10 039 Random effect associated to the tree Intercept 37.3 CA 7.76 Random effect associated to the disk Intercept 162 CA 5.34 CA 59.2 11+RW 325 Autocorrelation 0.393 Residuals 36.4 AIC 101434 BIC 101 572 Pseudo-R2 (%) Fixed 14.9 Random tree 30.8 Random disk 31.7 Total 77.4 Variables Parameter estimates Standard deviation Degrees of freedom Fixed Intercept 383 17.0 10 039 CA 0.237 0.518 10 039 CA 5.14 5.91 10 039 11+RW 196 23.9 10 039 Random effect associated to the tree Intercept 37.3 CA 7.76 Random effect associated to the disk Intercept 162 CA 5.34 CA 59.2 11+RW 325 Autocorrelation 0.393 Residuals 36.4 AIC 101434 BIC 101 572 Pseudo-R2 (%) Fixed 14.9 Random tree 30.8 Random disk 31.7 Total 77.4 Table 3 Summary statistics of model M1. Variables Parameter estimates Standard deviation Degrees of freedom Fixed Intercept 383 17.0 10 039 CA 0.237 0.518 10 039 CA 5.14 5.91 10 039 11+RW 196 23.9 10 039 Random effect associated to the tree Intercept 37.3 CA 7.76 Random effect associated to the disk Intercept 162 CA 5.34 CA 59.2 11+RW 325 Autocorrelation 0.393 Residuals 36.4 AIC 101434 BIC 101 572 Pseudo-R2 (%) Fixed 14.9 Random tree 30.8 Random disk 31.7 Total 77.4 Variables Parameter estimates Standard deviation Degrees of freedom Fixed Intercept 383 17.0 10 039 CA 0.237 0.518 10 039 CA 5.14 5.91 10 039 11+RW 196 23.9 10 039 Random effect associated to the tree Intercept 37.3 CA 7.76 Random effect associated to the disk Intercept 162 CA 5.34 CA 59.2 11+RW 325 Autocorrelation 0.393 Residuals 36.4 AIC 101434 BIC 101 572 Pseudo-R2 (%) Fixed 14.9 Random tree 30.8 Random disk 31.7 Total 77.4 Figure 6 View largeDownload slide Mean ring density as a function of (a) cambial age and (b) ring width. The solid lines correspond to a loess smoother. Figure 6 View largeDownload slide Mean ring density as a function of (a) cambial age and (b) ring width. The solid lines correspond to a loess smoother. When performing the forward stepwise regressions, the residuals of model M1, εM1, were found to be related to several climatic variables (Figure 7). When εM1 was related to all the climatic variables (model M2), the residual standard error of the model was 32.7 kg m−3 and the model accounted for only 8.0 per cent of the variation. The VIFs were checked and all ranged between 1 and 2 (not shown), leading to the conclusion that the level of multicollinearity in the model was acceptable. Overall, εM1 increased with decreasing temperatures of the fall and winter before the year of ring formation (previous October, January and February). In addition, spring temperatures (in April and May) were positively related to εM1, while summer temperatures (July and August) were negatively related to εM1. Other temperature variables were found to be significantly related to εM1 but no general trend could be inferred: June and August temperatures of the previous growing season were positively related to εM1, while it was the opposite for July temperatures of the previous growing season. When considering the effects of precipitation on εM1, results showed that monthly precipitation from June to August were positively related to εM1. The other effects of precipitation were unclear as there were no consecutive months for which significant effects had the same sign. Figure 7 View largeDownload slide Parameter values from the forward stepwise regression model of the residuals from M1 for (a) temperatures and (b) precipitation variables from January of the year previous to ring formation until October of the year of ring formation. All parameters had a P-value lower than 10−3, except August temperatures which had a P-value between 10−2 and 10−3. The letter p before the names of months refers to monthly climatic variables of the year previous to ring formation. The vertical dotted line separates climatic variablee of the year previous to ring formation and the year of ring formation. Figure 7 View largeDownload slide Parameter values from the forward stepwise regression model of the residuals from M1 for (a) temperatures and (b) precipitation variables from January of the year previous to ring formation until October of the year of ring formation. All parameters had a P-value lower than 10−3, except August temperatures which had a P-value between 10−2 and 10−3. The letter p before the names of months refers to monthly climatic variables of the year previous to ring formation. The vertical dotted line separates climatic variablee of the year previous to ring formation and the year of ring formation. The moving window Spearman’s rank correlation coefficient from model M1 showed that the correlation between observed MRD and predictions of model M1 was higher in the control stands than in the thinned stands for almost all the observation period (Figure 8a). The variations of the correlation coefficient were synchronous before thinning between thinned and control stands, but then diverged after thinning. Specifically, the correlation coefficient between MRD observations and predictions of the M1 model increased from 0.891 2 years before thinning to 0.923 3 years after thinning in control stands. Because the Gershunov test indicated that the average correlation coefficient was 0.897 with a confidence interval ranging from 0.883 to 0.911, this increase in the correlation coefficient significantly remained out its intrinsic range of variability for 5 consecutive years. Six years after thinning, the correlation coefficient decreased to within this intrinsic range of variability. In thinned stands, the correlation coefficient decreased from 0.879 2 years before thinning to a minimum of 0.847 6 years after thinning. In these stands, the Gershunov test indicated that the average correlation coefficient was 0.867 with a confidence interval ranging from 0.851 to 0.882. In this case, the correlation coefficient was lower than its intrinsic range of variability during a 4-year period, i.e. from 4 years after thinning to 7 years after thinning. Afterwards, the correlation coefficient increased to within the intrinsic range of variability. Figure 8 View largeDownload slide Moving-window Spearman’s rank correlation coefficients between (a) mean ring density observations and predictions of the M1 model and (b) residuals of the M1 model and predictions of M2. Solid and dashed lines correspond, respectively, to the local correlation computed for control and thinned stands on a moving window of 10 years with a 1-year lag. Each period is centred so that year i corresponds to the period (i − 4; i + 5). The light grey and dark grey envelopes correspond to the 90% bilateral confidence intervals of the moving-window correlations for control and thinned stands, respectively (Gershunov test, see section Materials and methods). Figure 8 View largeDownload slide Moving-window Spearman’s rank correlation coefficients between (a) mean ring density observations and predictions of the M1 model and (b) residuals of the M1 model and predictions of M2. Solid and dashed lines correspond, respectively, to the local correlation computed for control and thinned stands on a moving window of 10 years with a 1-year lag. Each period is centred so that year i corresponds to the period (i − 4; i + 5). The light grey and dark grey envelopes correspond to the 90% bilateral confidence intervals of the moving-window correlations for control and thinned stands, respectively (Gershunov test, see section Materials and methods). The moving window Spearman’s rank correlation coefficient between εM1 and residuals of model M2 tended to be higher in thinned stands than control stands for most of the time period considered (Figure 8b). Except for short periods (e.g. from 3 years before thinning until the year of thinning), the variations of the correlation coefficient were synchronous in both control and thinned stands. In control stands, the average moving-window correlation coefficient between εM1 and residuals of model M2 was 0.241, with a confidence interval ranging from 0.150 to 0.326. Except for the period extending from 33 to 30 years before thinning, the correlation coefficient remained within its intrinsic range of variability, even after thinning. For thinned stands, the average moving-window correlation coefficient was 0.309 with an intrinsic range of variability spanning from 0.231 to 0.378. The correlation coefficient was below the lower bound from 1 year before thinning to 3 years after thinning. This is concomitant with the short loss of synchronicity in the variations of the correlation coefficients between control and thinned stands. Discussion Effects of climatic variables on MRD Overall, temperatures of the previous winter, as well as precipitation of the previous year, had negative effects on εM1. This illustrates the importance of carry-over effects on wood formation (Barbaroux and Bréda, 2002; Kagawa et al., 2006), as already reported for trees growing at high altitude (e.g. Picea crassifolia Kom., Xu et al., 2012), boreal (e.g. P. mariana (Mill.) B.S.P., Xiang et al., 2014a), and temperate conditions (e.g. Picea Rubens Sarg., Conkey, 1979). Although the effect of temperatures of the previous winter on εM1 may appear curious, this may also be linked to a carry-over process (Barbaroux and Bréda, 2002). In the case of high winter precipitation, the following snowmelt in spring may increase water availability, which favours the production of cells with large lumen and may thus cause a decrease in εM1 (Fritts et al., 1991). The positive effect of spring temperatures on the mean ring density of black spruce trees was previously observed by Xiang et al. (2014a) and can be related to cambial phenology. Indeed, black spruce cambial activity is under strong influence of spring climatic conditions (Rossi et al., 2011), and is stimulated by favourable temperatures during the beginning of the growing season (Wang et al., 2002). Such observations have been generalized to all conifers growing in northern climatic conditions (Deslauriers and Morin, 2005; Rossi et al., 2008). These studies suggest that high spring temperatures may hasten the onset of cambial activity and stimulate both cell division and cell wall production in earlywood, therefore leading to higher earlywood and mean ring densities. The effect of current summer temperatures and precipitation is in accordance with the results of Xiang et al. (2014a). First, warm summers are often associated with high potential evapotranspiration that can reduce tree photosynthesis (Lebourgeois et al., 1998; Flexas et al., 2006), shorten the period of wood formation (Sohn et al., 2012) and ultimately decrease carbohydrate allocation to cell wall production. This would result in lower wood density, as evidenced in the present study. Second, we showed that summer precipitation was negatively associated with mean ring density. In the present study, precipitation was more abundant during the summer months, which likely increased water availability. Such high water availability is known to result in the production of cells with larger lumens (Fritts et al., 1991) as cell enlargement is still ongoing during latewood formation (Deslauriers et al., 2003). This ultimately leads to rings with lower wood density. Effects of thinning on MRD In this study, only a small effect of thinning on mean ring density was detected, while its effect on growth rate was generally positive. For numerous species, wood density decreases following thinning as a consequence of a higher growth rate (Pinus banksiana Lamb., Barbour et al., 1994; Schneider et al., 2008; Pinus sylvestris L., Peltola et al., 2007). However, in line with our results, the wood density response of black spruce is generally limited after partial cutting treatments such as careful logging around small merchantable stems (Lemay et al., 2016) and commercial thinning (Vincent et al., 2011). This can explain the decline in the strength of the relationship between mean ring density and growth rate that was evidenced by the correlation coefficient between predictions of model M1 and mean ring density observations. As inter-tree competition decreases following thinning, residual trees have better access to growth resources. In turn, thinned trees could use these resources to invest in secondary growth, which is a lower carbon allocation priority than height growth (Mccarthy and Enquist, 2007). This is consistent with our observation that the strength of the relationship between mean ring density and growth rate decreased after thinning. This potentially higher secondary growth production of thinned trees may be used to produce rings with higher proportion of earlywood (Wang et al., 2002; St-Germain and Krause, 2008) but also higher earlywood density, which would lead to a limited change in mean ring density. We were unfortunately unable to test this hypothesis because the slow growth rate of the sampled trees prevented us from differentiating the earlywood from the latewood (14 per cent of the sampled rings were narrower than 0.5 mm). It may also be that at such slow growth rates, an increase in growth rate leads to a similar increase of both the earlywood and latewood widths. Effect of thinning on the relationship between MRD and climatic variables The present study showed similar relationships between mean ring density and climatic variables for both control and thinned plots. MRD was under the influence of several climatic variables, similarly to the few previous studies conducted in temperate or boreal conditions (Wimmer and Grabner, 2000; Xiang et al., 2014b). Although these effects could be detected, they represented only a small proportion of the variation explained by the M2 model, which may explain why thinning did not influence the relationship between MRD and climatic variables. Such low influence of climatic variables on MRD may result from the presence of compression wood in discs. Specifically, compression wood may lead to unexplained variation in mean ring density that would be included in the residuals of the models. In addition, the small proportion of the MRD variation explained by climatic variables may result from the fact that we considered mean ring density. Latewood density is under a stronger climatic influence than earlywood density (Wimmer and Grabner, 2000) and both ring components respond differently to climatic variables (Franceschini et al., 2013b). Nonetheless, the limited dependence of MRD on climatic variables could also imply that some of the effects highlighted in the present study may have been detected, but they did not necessarily imply a causal relationship (Shipley, 2016). For example, the temperatures observed during June, July and August of the previous year were, respectively, associated with positive, negative and positive parameters in model M2 (Figure 7a). In addition, the climatic dependence of black spruce mean ring density in boreal conditions may not be strong enough to be modulated by thinning. Our sample trees were growing in conditions where climate has a lower influence on wood density than at the treeline (Gort-Oromi et al., 2011; Xiang et al., 2014a). While we observed rather large intra- and inter-annual variation in temperature, the growth conditions may not have been sufficiently limiting to affect MRD significantly. Intra-annual variations in precipitation showed that drought events are unlikely, and thus it is understandable that this climatic variable does not strongly influence the mean ring density. It is widely recognized that the annual ring characteristics of trees growing in harsh environments are strongly dependent on climatic variables. For example, summer temperatures are critical at the latitudinal treeline (Wang et al., 2002; Pritzkow et al., 2014), whereas tree ring indicators are influenced by summer droughts in the Mediterranean context (Büntgen et al., 2010; Rozas et al., 2011). Such strong climatic dependence of tree characteristics may reflect the fact that the growth limitation in harsh environments is mainly caused by the meristem activity (sink limitation) while trees growing away from the treeline are more limited by substrate availability (source limitation; Körner, 1998; Salzer et al., 2009). In addition, black spruce is a widespread species in North-America with a large ecological range (Burns and Honkala, 1990), which could reflect a lower sensitivity to external factors. Conclusion We found that mean ring density was unaffected by a slight increase in ring width following thinning in black spruce stands. This produced a decrease in the strength of the relationship between mean ring density and annual ring width. This was interpreted as an effect of better resource repartition in the stand. Further work is needed to consider temporal variations of photosynthetic capacity as a consequence of growth development and its effects on wood density. However, our second hypothesis was not confirmed, as no change in the effects of climatic variables on mean ring density was found. It is possible that (i) trees were not limited enough by climatic variables to exhibit such change and (ii) the large black spruce ecological niche makes the climatic dependence of MRD unaffected by thinning. Further testing of this hypothesis should be done both in temperate and in harsher climatic conditions. From a management point of view, the maintenance of black spruce wood density with increasing growth rate could lead to an increase in wood value after thinning. Although this would be attractive to both forest managers and wood processors, a full assessment of the economical returns of commercial thinning in these stands remains to be made. Acknowledgements The authors would like to thank Stephane Tremblay for giving access to the thinning trial data, Mikael Bernier and Marie-Pier Arsenault for field measurements, Mélanie Desrochers (CEF) for producing the site location map, Laurence Martel for wood density measurements, and the two anonymous reviewers and the associated editor for their helpful comments and suggestions on an earlier version of the manuscript. Funding This work was supported by the Fonds de Recherche du Québec – Nature et technologies and the Natural Sciences and Engineering Research Council of Canada Conflict of interest statement None declared. References Aussenac , G. and Granier , A. 1988 Effects of thinning on water stress and growth in Douglas-fir . Can. J. For. Res. 18 , 100 – 105 . doi:10.1139/x88-015 . Google Scholar CrossRef Search ADS Barbaroux , C. and Bréda , N. 2002 Contrasting distribution and seasonal dynamics of carbohydrate reserves in stem wood of adult ring-porous sessile oak and diffuse-porous beech trees . Tree Physiol. 22 , 1201 – 1210 . doi:10.1093/treephys/22.17.1201 . Google Scholar CrossRef Search ADS PubMed Barbour , R.J. , Fayle , D.C.F. , Chauret , G. , Cook , J. , Karsh , M.B. and Ran , S. 1994 Breast-height relative density and radial growth in mature jack pine (Pinus banksiana) for 38 years after thinning . Can. J. For. Res. 24 , 2439 – 2447 . doi:10.1139/x94-315 . Google Scholar CrossRef Search ADS Bontemps , J.-D. and Esper , J. 2011 Statistical modelling and RCS detrending methods provide similar estimates of long-term trend in radial growth of common beech in north-eastern France . Dendrochronologia 29 , 99 – 107 . doi:10.1016/j.dendro.2010.09.002 . Google Scholar CrossRef Search ADS Bouriaud , O. , Bréda , N. , Dupouey , J.-L. and Granier , A. 2005 Is ring width a reliable proxy for stem-biomass increment? A case study in European beech . Can. J. For. Res. 35 , 2920 – 2933 . doi:10.1139/x05-202 . Google Scholar CrossRef Search ADS Brazier , J.D. 1970 Timber improvementII. The effect of vigour on young-growth sitka spruce . For. Int. J. For. Res. 43 , 135 – 150 . doi:10.1093/forestry/43.2.135 . Bunn , A.G. 2008 A dendrochronology program library in R (dplR) . Dendrochronologia 26 , 115 – 124 . doi:10.1016/j.dendro.2008.01.002 . Google Scholar CrossRef Search ADS Büntgen , U. , Frank , D. , Trouet , V. and Esper , J. 2010 Diverse climate sensitivity of Mediterranean tree-ring width and density . Trees 24 , 261 – 273 . doi:10.1007/s00468-009-0396-y . Google Scholar CrossRef Search ADS Burns , R.M. and Honkala , B.H. , Technical C . 1990 . Silvics of North America. Volume 1. Conifers. vi + 675 pp. Bustos , C. , Beauregard , R. , Mohammad , M. and Hernandez , R.E. 2003 Structural performance of finger-jointed black spruce lumber with different joint configurations . For. Prod. J. 53 , 72 – 76 . Conkey , L.E. 1979 Response of tree-ring density to climate in Maine, U.S.A . Tree-Ring Bull. 39 , 29 – 38 . D’Amato , A.W. , Bradford , J.B. , Fraver , S. and Palik , B.J. 2013 Effects of thinning on drought vulnerability and climate response in north temperate forest ecosystems . Ecol. Appl. 23 , 1735 – 1742 . doi:10.1890/13-0677.1 . Google Scholar CrossRef Search ADS PubMed Deslauriers , A. and Morin , H. 2005 Intra-annual tracheid production in balsam fir stems and the effect of meteorological variables . Trees 19 , 402 – 408 . doi:10.1007/s00468-004-0398-8 . Google Scholar CrossRef Search ADS Deslauriers , A. , Morin , H. and Begin , Y. 2003 Cellular phenology of annual ring formation of Abies balsamea in the Quebec boreal forest (Canada) . Can. J. For. Res. 33 , 190 – 200 . doi:10.1139/x02-178 . Google Scholar CrossRef Search ADS Filipescu , C.N. , Lowell , E.C. , Koppenaal , R. and Mitchell , A.K. 2013 Modeling regional and climatic variation of wood density and ring width in intensively managed Douglas-fir . Can. J. For. Res. 44 , 220 – 229 . doi:10.1139/cjfr-2013-0275 . Google Scholar CrossRef Search ADS Flexas , J. , Bota , J. , Galmés , J. , Medrano , H. and Ribas-Carbó , M. 2006 Keeping a positive carbon balance under adverse conditions: responses of photosynthesis and respiration to water stress . Physiol. Plant. 127 , 343 – 352 . doi:10.1111/j.1399-3054.2006.00621.x . Google Scholar CrossRef Search ADS Fox , J. and Weisberg , S. 2011 An {R} Companion to Applied Regression . 2nd edn . Sage . Franceschini , T. , Bontemps , J.-D. and Leban , J.-M. 2012 Transient historical decrease in earlywood and latewood density and unstable sensitivity to summer temperature for Norway spruce in northeastern France . Can. J. For. Res. 42 , 219 – 226 . doi:10.1139/x11-182 . Google Scholar CrossRef Search ADS Franceschini , T. , Bontemps , J.-D. , Perez , V. and Leban , J.-M. 2013 a Divergence in latewood density response of Norway spruce to temperature is not resolved by enlarged sets of climatic predictors and their non-linearities . Agric. For. Meteorol. 180 , 132 – 141 . doi:10.1016/j.agrformet.2013.05.011 . Google Scholar CrossRef Search ADS Franceschini , T. , Longuetaud , F. , Bontemps , J.-D. , Bouriaud , O. , Caritey , B.-D. and Leban , J.-M. 2013 b Effect of ring width, cambial age, and climatic variables on the within-ring wood density profile of Norway spruce Picea abies (L.) Karst . Trees Struct. Funct. 27 , 913 – 925 . doi:10.1007/s00468-013-0844-6 . Google Scholar CrossRef Search ADS Fritts , H.C. , Vaganov , E.A. , Sviderskaya , I.V. and Shashkin , A.V. 1991 Climatic variation and tree-ring structure in conifers: empirical and mechanistic models of tree-ring width, number of cells, cell size, cell-wall thickness and wood density . Clim. Res. 1 , 97 – 116 . Google Scholar CrossRef Search ADS Gauthier , S. and Vaillancourt , M.-A. , 2008 . Aménagement écosystémique en forêt Boréale. PUQ. Gebhardt , T. , Häberle , K.-H. , Matyssek , R. , Schulz , C. and Ammer , C. 2014 The more, the better? Water relations of Norway spruce stands after progressive thinning . Agric. For. Meteorol. 197 , 235 – 243 . doi:10.1016/j.agrformet.2014.05.013 . Google Scholar CrossRef Search ADS Genet , A. , Auty , Da. , Achim , A. , Bernier , M. , Pothier , D. and Cogliastro , A. 2012 Consequences of faster growth for wood density in northern red oak (Quercus rubra Liebl.) . Forestry . doi:10.1093/forestry/cps057 . cps057. Gershunov , A. , Schneider , N. and Barnett , T. 2001 Low-frequency modulation of the ENSO–Indian monsoon rainfall relationship: signal or noise? J. Clim. 14 , 2486 – 2492 . doi:10.1175/1520-0442(2001)014<2486:LFMOTE>2.0.CO;2 . Google Scholar CrossRef Search ADS Gort-Oromi , J. , Mehtätalo , L. , Peltola , H. , Zubizarreta-Gerendiain , A. , Pulkkinen , P. and Venäläinen , A. 2011 Effects of spacing and genetic entry on radial growth and ring density development in Scots pine (Pinus sylvestris L.) . Ann. For. Sci. 68 , 1233 . doi:10.1007/s13595-011-0117-8 . Google Scholar CrossRef Search ADS Goudiaby , V. , Brais , S. , Berninger , F. and Schneider , R. 2012 Vertical patterns in specific volume increment along stems of dominant jack pine (Pinus banksiana) and black spruce (Picea mariana) after thinning . Can. J. For. Res. 42 , 733 – 748 . doi:10.1139/x2012-029 . Google Scholar CrossRef Search ADS Hacke , U.G. and Sperry , J.S. 2001 Functional and ecological xylem anatomy . Perspect. Plant Ecol. Evol. Syst. 4 , 97 – 115 . doi:10.1078/1433-8319-00017 . Google Scholar CrossRef Search ADS Hacke , U.G. , Sperry , J.S. , Pockman , W.T. , Davis , S.D. and McCulloh , K.A. 2001 Trends in wood density and structure are linked to prevention of xylem implosion by negative pressure . Oecologia 126 , 457 – 461 . doi:10.1007/s004420100628 . Google Scholar CrossRef Search ADS PubMed Hale , S.E. 2003 The effect of thinning intensity on the below-canopy light environment in a Sitka spruce plantation . For. Ecol. Manage. 179 , 341 – 349 . doi:10.1016/S0378-1127(02)00540-6 . Google Scholar CrossRef Search ADS Hughes , M.K. 2002 Dendrochronology in climatology – the state of the art . Dendrochronologia 20 , 95 – 116 . doi:10.1078/1125-7865-00011 . Google Scholar CrossRef Search ADS Jaakkola , T. , Mäkinen , H. and Saranpää , P. 2005 Wood density in Norway spruce: changes with thinning intensity and tree age . Can. J. For. Res. 35 , 1767 – 1778 . doi:10.1139/x05-118 . Google Scholar CrossRef Search ADS Kagawa , A. , Sugimoto , A. and Maximov , T.C. 2006 13CO2 pulse-labelling of photoassimilates reveals carbon allocation within and between tree rings . Plant Cell. Environ. 29 , 1571 – 1584 . doi:10.1111/j.1365-3040.2006.01533.x . Google Scholar CrossRef Search ADS PubMed Kantavichai , R. , Briggs , D. and Turnblom , E. 2010 Modeling effects of soil, climate, and silviculture on growth ring specific gravity of Douglas-fir on a drought-prone site in Western Washington . For. Ecol. Manage. 259 , 1085 – 1092 . doi:10.1016/j.foreco.2009.12.017 . Google Scholar CrossRef Search ADS Kennedy , R.W. 1995 Coniferous wood quality in the future: concerns and strategies . Wood Sci. Technol. 29 , 321 – 338 . doi:10.1007/BF00202581 . Google Scholar CrossRef Search ADS Kohler , M. , Sohn , J. , Nägele , G. and Bauhus , J. 2010 Can drought tolerance of Norway spruce (Picea abies (L.) Karst.) be increased through thinning? Eur. J. For. Res. 129 , 1109 – 1118 . doi:10.1007/s10342-010-0397-9 . Google Scholar CrossRef Search ADS Körner , C. 1998 A re-assessment of high elevation treeline positions and their explanation . Oecologia 115 , 445 – 459 . doi:10.1007/s004420050540 . Google Scholar CrossRef Search ADS PubMed Lachenbruch , B. , Moore , J.R. and Evans , R. 2011 Radial variation in wood structure and function in woody plants, and hypotheses for its occurrence. In Size- and Age-Related Changes in Tree Structure and Function, Tree Physiology . Meinzer F.C. , Lachenbruch B. and Dawson T.E. (eds). Springer , pp. 121 – 164 . Google Scholar CrossRef Search ADS Lebourgeois , F. , Lévy , G. , Aussenac , G. , Clerc , B. and Willm , F. 1998 Influence of soil drying on leaf water potential, photosynthesis, stomatal conductance and growth in two black pine varieties . Ann. Sci. For. 55 , 287 – 299 . doi:10.1051/forest:19980302 . Google Scholar CrossRef Search ADS Lemay , A. , Krause , C. , Achim , A. and Bégin , J. 2016 Growth and wood quality of black spruce and balsam fir following careful logging around small merchantable stems (CLASS) in the boreal forest of Quebec, Canada . For. Int. J. For. Res , 1 – 12 . doi:10.1093/forestry/cpw060 . Little , E.L.J. , 1971 . Atlas of United States trees, volume 1, conifers and important hardwoods: U.S. Department of Agriculture Miscellaneous Publication 1146, 9 p., 200 maps. Lohrasebi , H. , Mabee , W.E. and Roy , D.N. 1999 Chemistry and pulping feasibility of compression wood in black spruce . J. Wood Chem. Technol. 19 , 13 – 25 . doi:10.1080/02773819909349597 . Google Scholar CrossRef Search ADS Ma , S. , Concilio , A. , Oakley , B. , North , M. and Chen , J. 2010 Spatial variability in microclimate in a mixed-conifer forest before and after thinning and burning treatments . For. Ecol. Manage. 259 , 904 – 915 . doi:10.1016/j.foreco.2009.11.030 . Google Scholar CrossRef Search ADS Macdonald , E. and Hubert , J. 2002 A review of the effects of silviculture on timber quality of Sitka spruce . Forestry 75 , 107 – 138 . doi:10.1093/forestry/75.2.107 . Google Scholar CrossRef Search ADS Mccarthy , M.C. and Enquist , B.J. 2007 Consistency between an allometric approach and optimal partitioning theory in global patterns of plant biomass allocation . Funct. Ecol. 21 , 713 – 720 . doi:10.1111/j.1365-2435.2007.01276.x . Google Scholar CrossRef Search ADS Meinzer , F.C. , Lachenbruch , B. and Dawson , T.E. 2011 Size- and Age-Related Changes in Tree Structure and Function . Springer . Google Scholar CrossRef Search ADS Ministère des Ressources Naturelles 2015 Le guide sylvicole du Québec - Tome II: Les concepts et l’application de la sylviculture . Publications du Québec . Misson , L. , Nicault , A. and Guiot , J. 2003 Effects of different thinning intensities on drought response in Norway spruce (Picea abies (L.) Karst.) . For. Ecol. Manage. 183 , 47 – 60 . doi:10.1016/S0378-1127(03)00098-7 . Google Scholar CrossRef Search ADS MRNFP 2003 Manuel d’aménagement Forestier . 4e édn . Ministère des Ressources naturelles, de la Faune et des Parcs du Québec, Direction des programmes forestiers . ed. Parresol , B.R. 1999 Assessing tree and stand biomass: a review with examples and critical comparisons . For. Sci. 45 , 573 – 593 . Peltola , H. , Kilpeläinen , A. , Sauvala , K. , Räisänen , T. and Ikonen , V.-P. 2007 Effects of early thinning regime and tree status on the radial growth and wood density of Scots pine . Silva Fenn 41 , 489 – 505 . Google Scholar CrossRef Search ADS Pinheiro , J. , Bates , D. and DebRoy , S. 2007 . Linear and nonlinear mixed effects models. R Package Version 3, 57. Pinheiro , J.C. and Bates , D.M. 2000 Mixed-effects Models in S and S-PLUS . Springer . Google Scholar CrossRef Search ADS Pritzkow , C. , Heinrich , I. , Grudd , H. and Helle , G. 2014 Relationship between wood anatomy, tree-ring widths and wood density of Pinus sylvestris L. and climate at high latitudes in northern Sweden . Dendrochronologia . doi:10.1016/j.dendro.2014.07.003 . R Core Team , 2014 . R: A Language and Environment for Statistical Computing. Régnière , J. 1996 Generalized approach to landscape–wide seasonal forecasting with temperature–driven simulation models . Environ. Entomol. 25 , 869 – 881 . Google Scholar CrossRef Search ADS Régnière , J. and Bolstad , P. 1994 Statistical simulation of daily air temperature patterns eastern North America to forecast seasonal events in insect pest management . Environ. Entomol. 23 , 1368 – 1380 . Google Scholar CrossRef Search ADS Rossi , S. , Deslauriers , A. , Griçar , J. , Seo , J.-W. , Rathgeber , C.B. , Anfodillo , T. , et al. 2008 Critical temperatures for xylogenesis in conifers of cold climates . Glob. Ecol. Biogeogr. 17 , 696 – 707 . doi:10.1111/j.1466-8238.2008.00417.x . Google Scholar CrossRef Search ADS Rossi , S. , Morin , H. , Deslauriers , A. and Plourde , P.-Y. 2011 Predicting xylem phenology in black spruce under climate warming . Glob. Change Biol. 17 , 614 – 625 . doi:10.1111/j.1365-2486.2010.02191.x . Google Scholar CrossRef Search ADS Rozas , V. , García-González , I. and Zas , R. 2011 Climatic control of intra-annual wood density fluctuations of Pinus pinaster in NW Spain . Trees 25 , 443 – 453 . doi:10.1007/s00468-010-0519-5 . Google Scholar CrossRef Search ADS Salzer , M.W. , Hughes , M.K. , Bunn , A.G. and Kipfmueller , K.F. 2009 Recent unprecedented tree-ring growth in bristlecone pine at the highest elevations and possible causes . Proc. Natl. Acad. Sci. 106 , 20348 – 20353 . doi:10.1073/pnas.0903029106 . Google Scholar CrossRef Search ADS Schneider , R. , Zhang , S.Y. , Swift , D.E. , Bégin , J. and Lussier , J.-M. 2008 Predicting selected wood properties of jack pine following commercial thinning . Can. J. For. Res. 38 , 2030 – 2043 . doi:10.1139/X08-038 . Google Scholar CrossRef Search ADS Shipley , B. 2016 Cause and Correlation in Biology: A User’s Guide to Path Analysis, Structural Equations and Causal Inference with R . Cambridge University Press . Sohn , J.A. , Kohler , M. , Gessler , A. and Bauhus , J. 2012 Interactions of thinning and stem height on the drought response of radial stem growth and isotopic composition of Norway spruce (Picea abies) . Tree Physiol . doi:10.1093/treephys/tps077 . tps077. Soucy , M. , Lussier , J.-M. and Lavoie , L. 2012 Long-term effects of thinning on growth and yield of an upland black spruce stand . Can. J. For. Res. 42 , 1669 – 1677 . doi:10.1139/x2012-107 . Google Scholar CrossRef Search ADS St-Germain , J.-L. and Krause , C. 2008 Latitudinal variation in tree-ring and wood cell characteristics of Picea mariana across the continuous boreal forest in Quebec . Can. J. For. Res. 38 , 1397 – 1405 . doi:10.1139/X08-008 . Google Scholar CrossRef Search ADS Telewski , F.W. 1989 Structure and function of flexure wood in Abies fraseri . Tree Physiol. 5 , 113 – 121 . doi:10.1093/treephys/5.1.113 . Google Scholar CrossRef Search ADS PubMed Todaro , L. and Macchioni , N. 2010 Wood properties of young Douglas-fir in Southern Italy: results over a 12-year post-thinning period . Eur. J. For. Res. 130 , 251 – 261 . doi:10.1007/s10342-010-0425-9 . Google Scholar CrossRef Search ADS Tong , Q. , Tanguay , F. and Zhang , S.Y. 2011 Impact of commercial thinning on annual radial growth and wood density . Wood Fiber Sci. 43 , 311 – 325 . Vaganov , E.A. , Anchukaitis , K.J. and Evans , M.N. 2011 How well understood are the processes that create dendroclimatic records? A mechanistic model of the climatic control on conifer tree-ring growth dynamics. In Dendroclimatology, Developments in Paleoenvironmental Research . Hughes M.K. , Swetnam T.W. and Diaz H.F. (eds). Springer , pp. 37 – 75 . Vincent , M. , Krause , C. and Koubaa , A. 2011 Variation in black spruce (Picea mariana (Mill.) BSP) wood quality after thinning . Ann. For. Sci. 68 , 1115 – 1125 . doi:10.1007/s13595-011-0127-6 . Google Scholar CrossRef Search ADS Vincent , M. , Krause , C. and Zhang , S.Y. 2009 Radial growth response of black spruce roots and stems to commercial thinning in the boreal forest . Forestry . doi:10.1093/forestry/cpp025 cpp025. Wang , L. , Payette , S. and Bégin , Y. 2002 Relationships between anatomical and densitometric characteristics of black spruce and summer temperature at tree line in northern Quebec . Can. J. For. Res. 32 , 477 – 486 . doi:10.1139/x01-208 . Google Scholar CrossRef Search ADS Wang , L. , Payette , S. and Bégin , Y. 2001 1300-year tree-ring width and density series based on living, dead and subfossil black spruce at tree-line in Subarctic Quebec, Canada . The Holocene 11 , 333 – 341 . doi:10.1191/095968301674769686 . Google Scholar CrossRef Search ADS Wimmer , R. and Grabner , M. 2000 A comparison of tree-ring features in Picea abies as correlated with climate . IAWA J. 21 , 403 – 416 . doi:10.1163/22941932-90000256 . Google Scholar CrossRef Search ADS Xiang , W. , Auty , D. , Franceschini , T. , Leitch , M. and Achim , A. 2014 a Wood density-climate relationships are mediated by dominance class in black spruce (Picea mariana (Mill.) B.S.P.) . Forests 5 , 1163 – 1184 . doi:10.3390/f5061163 . Google Scholar CrossRef Search ADS Xiang , W. , Leitch , M. , Auty , D. , Duchateau , E. and Achim , A. 2014 b Radial trends in black spruce wood density can show an age- and growth-related decline . Ann. For. Sci. 71 , 603 – 615 . doi:10.1007/s13595-014-0363-7 . Google Scholar CrossRef Search ADS Xu , J. , Lu , J. , Bao , F. , Evans , R. and Downes , G.M. 2012 Climate response of cell characteristics in tree rings of Picea crassifolia . Holzforschung 67 , 217 – 225 . doi:10.1515/hf-2011-0144 . Zobel , B.J. and van Buijtenen , J.P. , 1989 . Wood Variation: Its Causes and Control. xv + 363 pp. © Institute of Chartered Foresters, 2017. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Forestry: An International Journal Of Forest ResearchOxford University Press

Published: Sep 23, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off