Table 1. Main characteristics of birch sap yield, trees and stands used in modelling the sap yield in the South Savo, North Karelia, and Northern Ostrobothnia regions in Finland. In total, 225 trees in 74 stands were tapped in 2019 and/or 2020 and measured by diameter. Not all stand characteristics were available for each stand. |
Variable | N | Mean | SD | Range |
Sap yield (litre tree–1) | 287 | 47.8 | 40.1 | 0.6–246.5 |
Tree diameter (cm) | 225 | 19.2 | 6.3 | 6.4–43.0 |
Mean height (m) | 72 | 16.0 | 4.1 | 6.0–25.0 |
Mean diameter (cm) | 65 | 18.3 | 4.4 | 11.0–35.0 |
Stand basal area (m2 ha–1) | 39 | 17.3 | 6.6 | 3.0–31.0 |
Number of stems (ha–1) | 53 | 933 | 520 | 340–2200 |
Stand age (years) | 66 | 40 | 18 | 10–90 |
| N | Rich (OMT) | Medium (MT) | Sub-dry (VT) |
Site fertility a | 73 | 41 (56%) | 28 (38%) | 4 (6%) |
Fig. 1. Birch sap yield tapped in two years in three regions in Finland as a function of tree diameter at breast height.
Table 2. The linear mixed model (Eq. 1) estimated for the seasonal sap yield of birch trees (litre tree–1). The fitting statistics are calculated using fixed effects. The modelling data consist of 281 observations from 221 trees in 72 stands tapped in 2019 and/or 2020. |
Variable a | Estimate | Std err. | t-value | p-value |
Intercept | –2.031 | 0.804 | –2.53 | 0.013 |
Year 2019 (ref. 2020) | 0.540 | 0.099 | 5.46 | <0.001 |
ln(d) (cm) | 1.089 | 0.160 | 6.82 | <0.001 |
ln(H) (m) | 0.729 | 0.265 | 2.75 | 0.007 |
Random effect | Variance | | | |
Stand (N = 72) | 0.248 | | | |
Tree (N = 221) | 0.083 | | | |
Residual (N = 281) | 0.394 | | | |
Fitting statistics | | | | |
R2 (%) | 55.2 | | | |
Bias (litre tree–1) | –5.69 | | | |
Bias% (%) | –10.7 | | | |
RMSE (litre tree–1) | 34.92 | | | |
RMSE% (%) | 65.4 | | | |
Fig. 2. Simulated development of the annual sap yield of the average tree (A and B) and the total sap yield ha–1 (C) in a planted silver birch stand on rich site (OMT) in Mikkeli, South Savo as a function of mean diameter or stand age. The stand development was simulated using the Motti simulator, and thinnings and final cutting were simulated according to the silvicultural recommendations.