Modeling and predicting solar radiation transmittance in mixed forests at a within-stand scale from tree species basal area

- Radiation under forest canopy was modelled in mono-specific and mixed stands.
- The Beer-Lambert law can be used in mixed stands with the basal area of each species.
- Extinction coefficients were not different between mixed and mono-specific stands.
- Transmittance can be predicted at local scale within a stand with local basal area.
- Predictions of the model are unbiased with a relative mean absolute error of 20%.

Light under tree canopy cover is essential for the study and understanding of plant diversity, regeneration, plant growth and many other forest ecosystem processes; however, quantifying light is difficult and requires specialized equipment. That is why proxies or models predicting light availability can help scientists to obtain estimates of transmittance and forest managers to better assess and adjust silvicultural practices at a reasonable cost. The main objective of our research was to develop a model to predict local solar radiation transmittance from species basal area in mono-specific and mixed stands of sessile oak (Quercus petraea) and Scots pine (Pinus sylvestris). Models based on the Beer-Lambert law were fitted and compared using 163 measures of solar radiation transmittance obtained from hemispherical photographs and light sensors. In mono-specific stands, local transmittance was predicted by local basal area considered on a radius equaling approximately tree height. In mixed stands, the same model broken down into its mono-specific components (local basal area of each species multiplied by their own extinction coefficients) predicted transmittance well. The extinction coefficients we obtained were very close to those previously established for these species and did not differ between mono-specific and mixed stands. Our model explained 77% of the variation in transmittance when random effects were included and 64% of the variance without taking into account these random effects. The predictive value of the model was good with high accuracy (mean signed deviation not significantly different from zero) and a fairly high precision (relative mean absolute error = 20%). The fact that tree canopy transmittance in mixed stands can be predicted by extinction coefficients obtained from mono-specific stands indicates that modifications in crown structure and leaf distribution are only slight, or even non-existent, when the two species grow in a mixture.

155