4.2.1. Initial Stand Age and Density
Age and density are essential factors in the forest structure, and influence the growth response; in our experimental treatments, we confirmed the decline in growth with age already observed in other studies [19
]. In younger and denser stands (80–100 years, 2300–2900 trees/ha), radial growth response doubled that in older and open stands (110–160 years, 1300–1900 trees/ha), thus confirming hypothesis 3 (Figure 5
). The growth response in older and open stands was lower and shorter, probably due to older trees with lower photosynthetic rates [59
] and shorter periods of cambial activity and xylem cell differentiation than younger trees [60
]. Older trees were also closer to their maximal height, leaving little room for vertical crown expansion after the release from lateral competition. Growth-age predictions in Thorpe et al. [22
] were similar to the results found in our study for older stands. However, our studied variants of shelterwood in younger stands had a higher growth response; the predictions, e.g., indicated an increase of 0.9 mm/year in 100 years old stands 9 years a.c., and we observed 0.7 and 1.6 mm/year for interior and edge trees in DS. This could be explained by the fact that their model did not consider the spatial position and GBC. Nonetheless, the age effect was less significant than treatment, position or GBC.
4.2.2. Silvicultural Treatment
The results demonstrated that all study treatments increased radial growth of residual trees. Contrary to our expectations, no significant differences were found between experimental shelterwoods and seed-trees, with the exception of CS and ST that showed a small significant difference. This difference can be explained by the different stand structure. In older stands, CS is the most effective treatment on radial growth and ST the least. Thus, treatments showed different growth responses for each stand structure (Figure 5
). According to the results of released trees, we consider that DS and MS are the best option to promote radial growth in younger stands and CS in older stands. However, in future research, we recommend studying the volume production and mortality at a stand level to assess if the growth responses of the residual trees are able to compensate the reduction in stand density by the partial cutting treatments.
Some minor differences in harvested intensity of our studied treatments were observed between older and younger stands (e.g., MS). This reflects the random variability one can expect from “real-life” mechanized operations with no tree or trail marking prior to the harvest. The causes of such variations are site topography that does not allow regularly spaced trails and different operators who select trees in the application of the silvicultural prescription. These elements are part of the experimental error, and are assumed as such.
4.2.3. Edge Effect
The edge effect created by skidding trails in partial harvests is one of the strongest effects measured in our study, and a subject little studied in boreal forests [61
]. To our knowledge, this is the first evaluation of the edge effect on radial growth after partial cutting in black spruce even-aged stands and one of the few studies with dendroecological data.
Our findings confirm that the edge effect of skidding trails on tree radial growth response is a complex phenomenon that interacts with many factors such as stand age and density, trail distribution within the treatments, mortality and tree social status. The stand structures showed different growth response in edge trees that varied depending on stand age and density [22
]. The results indicated more edge effect influence in younger stands, in accordance with Harper, et al. [61
]. In the case of older stands, the growth response of edge trees was similar to the results obtained by Genet and Pothier [29
] for black spruce and balsam fir mixed stands in old-growth irregular forest.
Different edge effect growth responses among the studied treatments could be explained by the fact that each treatment has a specific spatial pattern and, consequently, different edge surface and residual strip width (Table 1
and Figure 2
). From these results, we can expect that treatments with more edge surface would register higher augmentation in radial growth at the stand level, especially in younger stands where edge effect was greater. For instance, CS was the treatment with the least edge surface; we speculate that this could explain the lowest growth response in younger stands. However, CS had a greater response in older stands due to the low influence of edge effect.
In younger stands, the radial growth of edge trees was twice that of interior trees in DS, MS and ST confirming hypothesis 4. This result is in agreement with the findings in Pinus radiata
] and Tryplochiton scleroxylon
] stands that indicated a decrease of 50% in DBH values for interior trees. In Pinus contorta,
31% greater stand basal area was detected in edge trees between 3 and 15 years after road construction [37
]. In the case of Pinus taeda
and Liriodendron tulipifera
, differences of 5.2 and 8 cm have been identified between interior and edge trees 20 years after edge creation [62
]. Thus, it seems that the soil compaction and wounds to the roots and trunk on edge trees caused by machines during the cutting operations did not have a negative impact on growth response in the short term, as shown also in Picchio, et al. [63
]. This lack of impact may be related to the high ecological resilience to soil disturbances of this species, which occupies a wide spectrum of environments such as peatlands, permafrost soils, higher northern latitudes or mixed forest [64
], and grows at elevations ranging from sea level to 1500 m [65
]. Black spruce has the ability to endure stress situations like extreme water deficit [66
], and can develop adventive roots exceeding 2 meters (60% of total root length) in one year [67
Growth differences between edge and interior trees were correlated with the measured CIi
); this relationship has been reported in the literature [26
]. Edge trees in younger stands of DS showed the lowest competition values, and it was the studied shelterwood with the highest growth response. In DS, the numerous small gaps created by the combination of main and secondary trails and the tree selection inside the residual strip explain this situation. In ST, the creation of large gaps contributed to a comparable reduction in CIi
due to the high mortality a.c. of residual trees. Tree selection and mortality in the residual strip promoted the reduction in tree density and produced an increased canopy opening that favored the edge influence on residual trees [25
]. However, for interior trees in the same stand type, CIi
values for all the treatments were close to trees in control plots. The comparable value of CIi
between MS and CS suggests that tree-selection in CS was not sufficient to significantly reduce competition in comparison with a partial cutting without tree selection. On the contrary, tree selection influenced the smallest differences between edge and interior trees in CS. However, in DS, the growth response of interior trees was lower than CS, probably due to the residual strip being the widest in the studied treatments.
In older stands, the variability of CIi may be caused either by more heterogeneous initial tree distribution, or by random mortality that occurred after the partial cutting treatment (e.g., ST). Overall, relative differences in CIi between edge and interior trees and between treatments were less than for younger stands, which is correlated with the smaller growth response of older stands to the treatments. This could be explained by the fact that the same man-made gaps created in each treatment are proportionally less important in older stands than in younger ones, because of differing initial tree spacing and size.
The presence of a growth response even in interior trees that are not subjected to tree selection suggests that the depth of the edge effect probably extends close to 1.25 m from the trails, the distance that we arbitrarily chose for selecting edge trees. We speculate that the depth of the edge effect will be higher in older than younger stands due to less density, and in treatments with tree selection and high mortality a.c. In future investigations, measuring the tree distance from trails, as in Genet and Pothier [29
], could be added to our methodology for a more precise evaluation.
4.2.4. Time Response
The growth response was not immediate after treatment, the majority of trees showed a no response step (0–3 years a.c.) in agreement with previous researches [12
]. A possible explanation is resource allocation in the root system due to a stress response to new conditions a.c. (higher wind penetration, light intensity and transpiration) in order to promote stability, and uptake of water and nutrients [20
]. The growth response was delayed around 5–6 years, similar to the results found in other partial cutting studies [55
]. We speculate that the no response step and the cores extraction at breast height (1.3 m) influenced the delay time.
The temporal response in tree growth a.c. was affected by stand structure and tree positions in the residual strip. Our growth response in younger stands showed a peak 9 years a.c. then started to decrease. We hypothesize that the growth in younger stands continues to decrease gradually to the values shown b.c. [22
]. On the contrary, the growth peak in older stands was not obvious due to the high variability of trees; we thus assume that the radial growth would be stable for a few years before decreasing. Long-term monitoring is needed to confirm this. The response time in growth was 5 years in younger stands; this can be explained by windthrow disturbance in a younger control plot (the same year as cutting). We speculate that without this event, the response time would be close to 3 years a.c.
Different growth temporal responses were observed between position classes: Edge trees in MS and ST reacted one year before interior trees in younger stands (Figure 5
); the growth peak was the ninth year a.c. in MS and DS edge trees but interior trees continued to grow beyond that year (Figure 5
). This delay in temporal response could be explained by the edge trees having more accessibility to nutrients, higher soil temperature, lower competition, higher lateral light intensity than interior trees [25
], and reacted earlier. However, ST interior and edge trees experienced the growth peak in the same year. It is likely that the skidding trails area three times wider (15 m) than MS and DS, and the narrow residual strip (5 m) could affect the edge influence on growth response. In older stands, this delay in temporal response was not observed; we concluded that differences in the temporal response between edge and interior trees are not obvious in older stands.
4.2.5. Growth before Cutting
Our results suggest that the growth response of residual trees depends on GBC. The study demonstrated that suppressed trees show better growth ratio before and post-treatment than dominant trees, in agreement with other studies [15
]. Nonetheless, this phenomenon is influenced by structure, treatment and spatial position effects (Figure 7
). The response is amplified in younger and higher initial density stands in MS and DS treatments, notably for edge trees. This could be explained by suppressed trees experiencing more difficult growing conditions b.c. in high density stands, and the edge position decreases the competition for light and nutrients. Other factors that could influence the growth response in suppressed trees is the tree-selection, and mortality in the residual strip, especially in younger stands. Edge and suppressed trees in ST and CS of younger stands had slightly lower growth response than other treatments. This may be caused at least partly by greater drought stress or insolation from the large canopy openings or because ST was the silvicultural treatment with elevated mortality in our study (around 70% of trees), and edge trees with low DBH have high probability of death a.c. [30
]. Overall, growth response was stronger for dominant than for suppressed trees in absolute terms, thus not confirming hypothesis 5. This supports the hypothesis of asymmetric competition for light as the main process in the studied stands [74
]. For dominant trees on the edge of trails, DS and ST caused the strongest response, probably because of the elimination of a greater number of competitors in the immediate surroundings of the residual trees in comparison with other treatments.