Half-century evidence from western Canada shows forest dynamics are primarily driven by competition followed by climate

Abstract

Tree mortality, growth, and recruitment are essential components of forest dynamics and resiliency, for which there is great concern as climate change progresses at high latitudes. Tree mortality has been observed to increase over the past decades in many regions, but the causes of this increase are not well understood, and we know even less about long-term changes in growth and recruitment rates. Using a dataset of long-term (1958-2009) observations on 1,680 permanent sample plots from undisturbed natural forests in western Canada, we found that tree demographic rates have changed markedly over the last five decades. We observed a widespread, significant increase in tree mortality, a significant decrease in tree growth, and a similar but weaker trend of decreasing recruitment. However, these changes varied widely across tree size, forest age, ecozones, and species. We found that competition was the primary factor causing the long-term changes in tree mortality, growth, and recruitment. Regional climate had a weaker yet still significant effect on tree mortality, but little effect on tree growth and recruitment. This finding suggests that internal community-level processes-more so than external climatic factors-are driving forest dynamics.

Keywords: boreal forest; climate change; forest dynamics; tree competition; tree demographic rates.

Fig. 1.

Locations of 1,680 permanent sample plots (PSPs) in western Canada. Each dot stands for one PSP. Four colors (dark blue, red, pink, and light blue) were used to show the distribution of PSPs in each of four provinces: British Columbia (777), Alberta (563), Saskatchewan (290), and Manitoba (50). The background colors represent Canada’s ecozones.

Fig. 2.

Modeled trends (1958–2009) of tree mortality (A and B), growth (C and D), and recruitment (E and F) rates at both stand and species levels. A generalized nonlinear mixed model was used for modeling temporal trends of tree mortality and recruitment rates, and a linear mixed model was used for modeling temporal trends of tree growth rates (Methods). At the stand level, PSPs were classified into three age groups based on initial stand ages: young forests (50–80 y), mature forests (80–120 y), and old-growth forests (≥120 y). At the species level, modeled trends of nine tree species are shown, including, AW, trembling aspen (Populus tremuloides); BW, white birch (Birch papyrifera); FB, balsam fir (Abies balsamea); FD, Douglas fir (Pseudotsuga menziesii); PB, balsam poplar (Populus balsamifera); PL, lodgepole pine (Pinus contorta); PJ, jack pine (Pinus banksiana); SB, black spruce (Picea mariana); and SW, white spruce (Picea glauca).

Fig. 3.

Summary of the relationships among tree demographic rates, competition, and climate at both stand and species levels. Each bar in the bar plots in A–C shows the percentage of the regression models (among all of the models tested) in which each of the variables (on the x label) was positively significant, negatively significant, or nonsignificant. We divided the 1,680 PSP data into different groups, including stand age groups, provinces, ecozones, elevations, and tree size. Within each group, the plots were stratified into different classes (e.g., different age classes, different provinces, etc.) and the data of each class were fitted using GNMM for tree mortality and recruitment and LMM for tree growth (Methods). Those group classes with <20 plots were excluded from modeling. The competition indexes (BA, stand basal area; BAL, basal area of larger trees; SDI, stand density index) were entered into the models separately, and the three climate variables (MWMT, mean warmest month temperature; MCMT, mean coldest month temperature; MAP, mean annual precipitation) were included in the models simultaneously. Three colors in these bar plots represent three significant levels: significantly positive relationships (dark brown), significantly negative relationships (light brown), and nonsignificant relationships (greenish brown). The box plots in D–F show the standardized regression coefficients between demographic rates and competition indexes and climate variables at the stand level. The box plots in G–I show the standardized regression coefficients between the demographic rates and the competition indexes and climate variables for three selected tree species (Methods). Only the result for the competition index BA is shown as the two other competition indexes had similar results.