This study was carried out in the Champagne and Aishihik Traditional Territory (CATT) within SWY during years between 2008 and 2012. The CATT extends across four eco-regions of Yukon: the St. Elias Mountains, the Ruby Range, the Yukon Southern Lakes and the Yukon Stikine Plateau, representing a landscape of great diversity and variability. The study area covers an altitudinal range between 60.63˚N and 61.2˚N and longitudinal range between −137.35˚W and −136.97˚W (Figure 1). The elevation range of the study area is from 651 to 1158 m above Mean Sea Level (MSL). The climate of the study area is characterized by cold winters and warm and dry summers with mean annual temperature −1.65˚C andmean annual precipitation of 319.5 mm. Soil texture varies from mixtures of sandy and silty fluvial deposits by rivers and lakes in lowland areas to mixtures of sandy and silty clays in the rest of the area. Permafrost is also common but discontinuous in the study area.

The study area is dominated by white spruce (Picea glauca, (Moench) Voss), followed by trembling aspen (Populus tremuloides Michx.) and balsam poplar (Populus balsamifera L.). The majority of the stands are between 80 and 120 years-old, although about 20% of the stands are younger than 70 years-old (Government of Yukon, 2004) . Sub-alpine and alpine plant communities are abundant in the mountainous areas above 1100 meters.

The majority of the study area has been disturbed by a spruce bark beetle (Dendroctonus rufipennis) infestation and by forest fires. Spruce bark beetles reached epidemic levels in the region after 1992, subsiding in 2007. Approximately 340,000 hectares of forest were affected by the beetle outbreak (NCE, 2006; Government of Yukon, 2010) . Similarly, fire is common in the study area. Between 1946 and 2003, the region experienced 23 larger fires and numerous smaller fires (Government of Yukon, 2004) . Most areas in the region were burned in 1950s.

Stratified systematic sampling (Kindt and Coe, 2005) was applied for the collection of data. Ecological, topographical and edaphic data were collected from 90 plots located along eight transects within the forested landscape of the CATT (Figure 1) in the summer of 2008.The plots were classified into five strata based on the disturbance type: a) undisturbed forest (control); b) an old fire-affected forest (more than 60 years-old with no trace of beetle); c) a recent fire-affected forest (less than 15 years-old and previously affected by beetle); d) a beetle- affected forest; and e) a salvage harvesting area (previously affected by beetle). The disturbed plots were identified and mapped using the local knowledge of the Yukon Government’s Ministry of Energy, Mines and Resources (MEMR) and the Champagne and Aishihik First Nation (CAFN) government. The plots were also classified according to disturbance severity as follows: 1) high severity: disturbed by a combination of either spruce bark beetle and fire (fire plots) or spruce bark beetle and salvage harvesting (harvesting plots); 2) medium severity: disturbed more than 50% by spruce bark beetle infestation without fire and salvage harvesting (beetle plots); and 3) low severity plots: signs of old fire and not affected by spruce bark beetle (old fire plots; 1948 constitutes the earliest available data for this region, hence, older fires are grouped here).

For each transect, a point of commencement (POC) was chosen, which was generally a visible landmark nearby an access point. The first plot was selected 100 m from the POC along a random bearing. Subsequent plots were systematically located 100 m from the previous plot. At each plot location, a 20 m × 20 m (400 m²) plot was established to collect elevation, slope, aspect, latitude, longitude, tree density, and crown coverage data. Within each 400 m² plot, three 25 m² nested sub-plots were randomly selected to collect shrub and sapling data. Seedling density and herb, bryophyte, and lichen cover were collected separately from three additional randomly selected 1 m × 1 m micro-plotslocated within the 25 m² subplots. One soil pit was located in a randomly selected micro-plot for soil sampling and characterisation.

Average height, diameter and basal area were calculated by species and dominance class. Density, total crown cover, and total basal area were calculated by disturbance type. Stand structure and composition was assessed using density index. Proportion (%) of total density (number of stems/ha) comprised of trees, seedlings, and saplings was calculated by species. Species composition was assessed using basal area ratio of a species with the total basal area (Huang and Titus, 1995) . The value ranges from 0 to 1 where value 0 indicates the absence of the species and 1 indicates a single species stand. We developed a density index to compare the density of trees, saplings and seedlings by species and disturbance regime with respect to average density on all 90 plots as follows;

where D_i is the density index, D_d is the average density of tree, seedling or sapling in a particular disturbance regime and D_t = average density of tree, sapling or seedling in 90 plots.

R² values were calculated for the variables related to stand structure. Scatter plots with regression fit line were made when both dependent and independent variables are numerical and ANOVA was used for the nominal independent variables. The significant of the plots were tested for 95% confidence level (P < .05). R software was used to calculate ANOVA, R squared values and to plot the relationship among variables (R Development core team, 2012) . In addition to these statistical tools, Multiple Co-inertia Analysis (MCA) was used to assess the relationship between environmental variables and stand structure (Dray et al., 2003) . ArcGIS 9.1 was used to illustrate the species distribution, crown cover and forest types in the study area (ESRI, 2009) .

We sampled a total of 2944 trees in 90 plots. White spruce dominates the landscape with 87.5% stand coverage followed by aspen and balsam poplar, which cover 12.2% and 0.3%, respectively (Table 1). Mixed stands had higher stand density (1063 stems/ha) than pure aspen stands (300 stems/ha) and white spruce stands (848 stems/ha). Similarly, white spruce had the highest sapling cover with 67.8%, followed by aspen (31.4%) and balsam poplar (2.6%) (Table 1). Stand density index analysis indicated that white spruce has the highest index (0.907) indicating the highest abundance on the landscape followed by aspen (0.09) and balsam poplar (0.001) (Table 2).

Tree size vs stand density: We found an inverse relationship between tree size and stand density for spruce, in agreement with our first hypothesis that diameter declines with increasing density. The density of co-dominant white spruce was 4% less than that of the intermediate class, but the basal area was 22% higher (Table 3), indicating that there is an inverse relationship between tree size and density of the stand and that the average diameter plays a greater role than density in increased forest productivity (basal area). The results also indicated that a decreasing trend in the average diameter of white spruce occurs when tree density exceeds 1000 stems per hectare (SPH). Generally, stand density has an inverse relationship with tree size (Lee et al., 1997; Delong and Kessler, 2000; Martin and Gower, 2006) , resulting in old growth stands having larger trees and lower densities due to self-thinning processes (Delong and Kessler, 2000) . Our results suggest that this relationship holds true below 1000 SPH; at greater stand densities tree size may become limited by inter-specific competition, whereas at lower densities other factors are at play (Huang and Titus, 1995) . The results suggest that thinning and selective harvesting in dense stands could play an important role in increasing forest productivity in the region.

White spruce age vs. size: The ages of white spruce were correlated with height, diameter, and basal area. However, McCarthy and Weetman (2006) reported from their study in Newfoundland that the age difference of trees with equal height and diameter could be more than 120 years. In our study, trees of similar size were found to be >120 years apart, indicating that tree size is a poor indicator for defining old growth stands.

Stand type vs. species basal area and density: Our finding of higher basal area in mixed stands than in pure white spruce stands contrasts with our expectation that productivity would be higher in pure than mixed stands. Linder (1998) also reported similar findings in his study in central Sweden, namely that the basal area was more than 21% higher in mixed stands compared to pure coniferous stands. In this regard, Kabzem et al. (2007) suggest that the presence of a small amount of aspen has a large effect on spruce productivity but that this effect diminishes when the aspen density is greater than 1000 stems per hectare. However, Stewart et al. (2003) reported that the basal area of pure coniferous stands was higher compared to mixed wood stands. Similarly, Hély et al. (2000) reported higher basal areas in deciduous forests followed by coniferous and mixed stands, but Linder (1998) and Stewart et al. (2003) found lower BA in deciduous stands than old mixed woods or coniferous stands. The reason for the higher basal areas of mixed wood stands in this study may be due to the presence of higher stand densities in mixed stands compared to pure stands in the study area.

Stand height and diameter: In this study average heights and diameters of both trembling aspen and white spruce were higher at sites at lower elevation, and on lower slope positions, agreeing with the expectations in our first hypothesis. Site productivity has been suggested as one of the main factors influencing the diameter structure of coniferous stands (Boucher et al., 2006) . As soil properties can affect the height growth of boreal trees (Martin and Gower, 2006) , lower rates of decomposition and high rates of leaching can cause lower nutrient availability at higher elevations and on steep slopes (Vitousek et al., 1988; Cole, 1995) , and these factors may have affected the lower height and diameter growth in higher elevations in this study.

White spruce’s growth responded positively to soil moisture. Soil moisture has been reported to be one of the important factors affecting site index of white spruce (Wang, 1995) ; it also controls the southern limit of white spruce in western Canada (Chhin and Wang, 2008) . Similarly, higher average spruce diameters and heights were found on NE and south aspects and the lowest on N or SE aspects. The reduced diameter growth on the N aspect can be explained by lower temperature, reduced availability of light and thereby shorter growing seasons. Conversely, the reduced diameter on SE aspects could be due to temperature-induced drought (Barber et al., 2000; Juday et al., 2003) . Generally hot, dry, summer conditions are exacerbated on S and SE aspects, and may reduce radial growth (Chapin et al., 2004; Barber et al., 2004) . Temperature-induced summer droughts in the western boreal region were recorded by Barber et al. (2000) .

Stand basal area: We found that forest productivity (basal area per hectare) varied significantly by geographical position, with higher basal area at lower elevations. This could have important implications for prescribing site-specific silvicultural systems in the study area. A similar trend was found in interior and northern British Columbia (Klinka et al., 1996) , and north-western Quebec (Chen et al., 1998 and Boucher et al., 2006) . White spruce basal area was higher in older stands consistence with the findings of Popadiouk et al. (2003) , who reported increasing basal area with stand age in eastern boreal mixed woods.

In this study, disturbances had significant effects on forest composition including species densities and their regeneration confirming our second hypothesis. Disturbances had promoted the development of a broadleaved forest in the region. A higher density of white spruce regeneration was found in plots with some kind of past disturbance such as fire or spruce bark beetle, meeting our expectation that white spruce regeneration density would increase with disturbance. The higher regeneration of trembling aspen in recently harvested fire plots indicated that fire promotes aspen regeneration and that the gap dynamics created by these disturbances (harvested plots had 29% mean crown coverage) were sufficient to promote the regeneration of this light-demanding species (Cumming et al., 2000; Kuusela, 1990) . Similarly, balsam poplar and its regeneration were also higher in disturbed forests, particularly in harvested plots. Balsam poplar, like trembling aspen, is a shade-intolerant species and prefers open areas to regenerate and grow (Alexander et al., 1990) . Although disturbances favour regeneration of all three species, the increased proportion of broadleaves following disturbance met our expectation that the proportion of broadleaved trees and regeneration would increase with disturbance.

We partially accept the third research hypothesis, that one or more climatic variables (e.g. MAT, MAP, TMXsm and NFFD) significantly affect the stand structure of all three species. We found higher basal area of white spruce on sites with a higher accumulation of growing degree-days and higher soil moisture. Similarly, the results of the multivariate analysis suggested that average diameter increases with increasing mean annual precipitation. These two results are consistent with our expectation that stand productivity would be greater on warmer and wetter locations. This expands the understanding of Wang and Klinka (1996) in their study in British Columbia, who suggested that the site productivity of white spruce was mostly influenced by edaphic factors rather than by climate. In contrast, Chavardes et al. (2012) , who undertook their study in the same research plots as our study, found that precipitation was important in determining the growth of white spruce, although the relationship varied month to month, with a negative relation found with April precipitation. Barber et al. (2000) reported that the radial growth of white spruce has decreased with increasing temperature in their study sites in Alaska, although this could be connected to soil moisture. Chavardes et al. (2012) also indicated that although white spruce growth was negatively related to maximum temperatures in general, increased precipitation accompanied by warmer temperature were correlated with increased tree growth in the last 30 years. It seems feasible that precipitation is a key factor affecting tree growth in the area but that it may interact negatively with increasing temperature. The mean annual precipitation in the study area is 319.5 mm (Wang et al., 2006) , which is projected to increase by 20% (ACIA, 2004) , which may favour spruce growth if temperature is not a limiting factor.The multivariate analysis used in this study suggested that the average height of white spruce increased with increasing summer maximum mean temperature. Climatic factors are considered more important than edaphic factors for white spruce height growth as height variations have been found in different climatic regions with the same edaphic site index (Wang et al., 1994) . However, it has also been reported that summer drought caused by increased temperature and hence evapo-transpiration may limit the growth of white spruce (Barber et al., 2000) . A similar result was found by Wilmking et al. (2004) in their study in Alaska.