Data

Data Exploration

This study examines the effects of burned legacy structures on soil moisture and seedling regeneration by measuring rainfall partitioning and seedling responses across seven treatment types: Open Control Group (OCG), Dense Snag Plot Center (DSPC), Dense Snag without Redirected Stemflow (DSw/oRS), Dense Snag with Redirected Stemflow (DSwRS), Individual Snag Plot Center (ISPC), Individual Snag with Redirected Stemflow (ISwRS), and Individual Snag without Redirected Stemflow (ISw/oRS).

Predictor variables include treatment type, total rainfall, throughfall, stemflow, and soil moisture (when analyzed as a factor influencing seedling growth). Response variables include soil moisture (when influenced by rainfall and treatment type), seedling height, total leaf area, and leaf water content. All variables were quantitatively measured, allowing for an analysis of how microclimatic conditions created snags influence soil water availability and seedling growth in a burned subalpine forests.

Seedling Growth:

Aspen seedlings were planted on May 17, 2024. Before planting, we measured a subsample to determine the average height and selected those between 20–30 cm. Initial height was recorded immediately after planting, and final measurements (height, shoot length, leaf area, and relative water content) were taken on September 18 (Table 1). Our predictor variable is treatment type, and our response variables are seedling growth and leaf area.

Dormant seedlings, which showed no growth, were excluded to avoid skewing results (Figures 1 & 3). This allowed for a clearer assessment of treatment effects on actively growing seedlings. Figures 1 and 3 show boxplots of seedling growth and leaf area across treatment types, respectively. Corresponding histograms of residuals (Figures 2 & 4) were used to check the normality assumption for ANOVA. Although some deviation from normality was visible, the Shapiro-Wilk test on model residuals—appropriate for ANOVA—indicated no serious violations. As a result, one-way ANOVA was used for both response variables.

Table 1. Shortened version of data sheet for seedling growth, leaf area, and relative water content. This data only represents a couple of each treatment type, and only 1 (of 3) of the seedlings that is planted around each snag.

Figure 1. Seedling growth (cm) across treatment types. OCG (n = 37) showed lower growth compared to DSwRS (n = 33), DSw/oRS (n = 36), and ISw/oRS (n = 6). Other treatments: DSPC (n = 28), ISPC (n = 17), ISwRS (n = 13).

Figure 2. Histogram of residuals from the one-way ANOVA model comparing seedling leaf area across treatment types. Residuals are centered near zero but deviate slightly from a normal distribution, as indicated by visual inspection and a Shapiro-Wilk test (p < 0.001). ANOVA was still considered appropriate due to balanced group sizes and absence of extreme outliers.

Figure 3. Boxplot of seedling leaf area (cm²) across treatment types. OCG (n = 37) showed lower leaf area compared to DSwRS (n = 33), DSw/oRS (n = 36), and ISw/oRS (n = 6). Other treatments: DSPC (n = 28), ISPC (n = 17), ISwRS (n = 13).

Figure 4. Histogram of residuals from the one-way ANOVA model comparing seedling leaf area across treatment types. Residuals were approximately normally distributed and centered near zero. A Shapiro-Wilk test confirmed no significant deviation from normality (p = 0.8993), meeting the assumptions of ANOVA.

Soil moisture:

Soil moisture was measured throughout the growing season using TOMST TMS-4 sensors, which recorded volumetric water content (VWC, %) at 15-minute intervals. To summarize sensor data across treatment types, average VWC was calculated for each sensor and grouped accordingly. Outliers were removed using the IQR method to reduce the influence of extreme weather events or potential sensor interference, allowing for a clearer comparison of treatment effects.

After cleaning, the data were visually and statistically assessed for normality. A histogram of residuals (Figure 6) showed a roughly symmetric distribution, and a Shapiro-Wilk test confirmed no serious violation of the normality assumption (p = 0.8395). Based on these results, a one-way ANOVA was used to compare soil moisture among treatment types.

Table 2. This table is showing the data that comes form the TOMST TMS-4 microclimate extreme monitoring station.

Figure 5. Growing season soil moisture (VWC %) by treatment type with outliers removed using the IQR method. Treatments include OCG (n = 15), DSPC (n = 16), DSwRS (n = 44), DSw/oRS (n = 44), ISPC (n = 4), ISwRS (n = 17), and ISw/oRS (n = 5).

Figure 6. Histogram of residuals from the one-way ANOVA model comparing soil moisture (VWC %) across treatment types. Residuals appeared normally distributed, confirmed by a Shapiro-Wilk test (p = 0.8395).

Stemflow and Throughfall:

Rainfall events were defined as periods of precipitation separated by at least six hours without rain, following Brasil et al. (2022). Total rainfall, throughfall, and stemflow were measured in liters per square meter (L/m²), with sample values shown in Table 3. Outliers caused by battery loss, wildlife disturbance, or extreme weather were removed, and gauges with missing data were excluded. Total rainfall was calculated using four open-area gauges. Throughfall and stemflow were then analyzed across treatment types and rain event sizes.

Figures 7 and 9 show stemflow and throughfall responses to rain event size. Residual plots (Figures 8 and 10) and Shapiro-Wilk tests showed a deviation from normality (p < 0.001) for stemflow, but one-way ANOVA was still applied given large, balanced sample sizes and the absence of extreme outliers.

Table 3. Rain event data. This table is showing a small portion of the rainfall, stemflow, and throughfall data is like.

Figure 7. Mean stemflow (L/m²) by rain event size for dense stand snags and individual snags. Events: small (≤5 L/m²), medium (5–14 L/m²), large (≥32 L/m²). Error bars: SE. Sample sizes (dense, individual): small (377, 174), medium (169, 78), large (52, 24). Individual snags showed higher stemflow.

Figure 8. Histogram of residuals from the one-way ANOVA model assessing stemflow across rain event sizes. Residuals cluster near zero but deviate from a normal distribution. Shapiro-Wilk test indicates significant deviation from normality (p < 0.001). ANOVA was still considered appropriate due to the robustness of the test with large samples and lack of extreme outliers.

Figure 9. Mean throughfall (L/m²) by rain event size for dense stands and open areas near snags. Events: small (≤5), medium (5–14), large (≥32). Error bars: SE. Sample sizes (dense/open): 377/232 (small), 169/104 (medium), 52/32 (large). Throughfall increased with event size, showing similar trends across groups.

Figure 10. Histogram of residuals from the one-way ANOVA model assessing throughfall across rain event sizes. Residuals are tightly centered around zero but are not perfectly normal. A Shapiro-Wilk test confirmed deviation from normality (p < 0.001), but one-way ANOVA remained suitable due to the absence of outliers and well-distributed sample sizes.

Page updated

Report abuse