Data

Initially, I recorded different information related to geographic position of the sampling plot and the characteristics of the line (table 4). I used this metadata to check possible errors in my main dataset (e.g., outliers in the response variable, erroneous levels in the predictor variables).

Table 4: Metadata associated to the observations.

Table 5 displays an abbreviated version of my main dataset. Responses variables are represented by the column "ABUNDANCE" that reports the number of stems counted in each m2 plot, and column "ABUNDANCE_HA" that reports the number of stems per hectare (multiplying the number of stems counted in one 20 m2 plot by 500). The main predictor categorical variable used for my analyses is the column "TREATMENTS" which include the combination of disturbances associated to each observation. Note that this variable is also presented through two binomial variables “HARVESTING" and "LINE" (presence/absence of cut-blocks and lines) for the Generalized Mixed models. Other categorical variables are "SPECIES”, “GROUP” and "HEIGHT" used to group my data based on the experimental question (see methods section). Finally, "AREA_ID", "SITE_ID" and "PLOT_ID" represent my design variable used to connect my observations with the related metadata and for the random effects in the GLMM, as factorial variables.

Table 5: Main dataset used for the analyses.

Exploratory graphics

Next graphs show the raw data sampled with the regeneration survey and used for the analyses. Initially, I checked for the distribution of the total stem density (independently by the species) in the different disturbance scenarios and height categories per 20 m2 plot (figure 8). Secondly, the same variables is presented as different boxplots for each disturbance scenario and height to present the quantiles and the range of observed values (figure 9). Values of total stem denisty for each plot is obtained by the sum of all rows associated with same PLOT_ID (i.e., 1 row for each plot and 15 rows for each TREATMENTS level). Analogues boxplots were made for each species group (figure 10) and single species (figure 11), but only for height category 2. Values of stem denisty for each plot and species group is obtained by the sum of all rows with the same PLOT_ID and the same value of variable GROUP (i.e., 3 row for each plot and 45 rows for each TREATMENTS level). Values of stem denisty for each species per plot is obtained by the sum of all rows with the same PLOT_ID and the same value of variable SPECIES (i.e., 6 row for each plot and 90 rows for each TREATMENTS level). All plots are made with the "ggplot2" package on R version 4.2.0.

Figure 8: Frequency classes of the total stem density for the different disturbances and height categories. Each bar represents the number of observations associated with the total stem density values in a single plot.

Figure 9: Quantiles, ranges and outliers of total stem densities for different disturbance scenarios and each height category.

Preliminary data exploration showed substantial differences in total stem densities between treatments and height categories. Fore hieght category 2, lots in the adjacent forest inside cut-blocks showed the highest average total stem density of 26 n/20 m2 (ranging from 0 to 71 n/20 m2). For the same height category, seismic lines in cut-blocks present an average stems density of 19 n/20 m2 (ranging from 0 to 52 n/20 m2). For height category 1 the differences between different treatments are not as substantiall as for height category 2.

Figure 10: quantiles, ranges and outliers of the stem density for different disturbance scenarios and for each species group.

Data exploration for different species groups showed that deciduous species represented the most common plant species in cutblok plots. More precisely, I found a mean density of 25 n/20 m2 in adjacent forest plots (ranging from 0 to 71 /20 m2) and 17 n/20 m2 in seismic lines plots (ranging from 0 to 52 /20 m2). Shrubs species represent the second most abundant group with a mean stem density of 3 n/20 m2 (ranging from 0 to 10 n/20 m2) and 3 n/20 m2 (ranging from 1 to 16 n/20 m2) respectively for plots in seismic lines and plots in adjacent forest inside cutblocks. Coniferous showed the lowest stems densities for all treatments with no substantial differences between distrubance scenarios.

Figure 11: Quantiles, ranges and outliers of the stem density for different disturbance scenarios and for each single species. The species presented in the graph are Trembling aspen (Aw), Balsam poplar (Bp), Balsam fir (Fb), Paper birch (Pb), Willow spp. (Salx) and White spruce (Sw).

Overall data exploration shows a very low abundance per plot for most of the sampled species, except for aspen (Aw), willows spp. (Salx) and paper birch (Bp). Specifically, for aspen I found an average stems density of 25 n/m2 (ranging from 0 to 71 n/20 m2) and 13 n/20 m2 (ranging from 0 to 40 n/20 m2) respectively for adjacent forest and seismic lines in cut-blocks. For willows spp. I found an average stems density of 4 n/m2 (ranging from 0 to 16 n/20 m2) and 3 n/20 m2 (ranging from 0 to 10 n/20 m2) respectively for adjacent forest and seismic lines in cut-blocks.