Methods for predicting Sitka spruce natural regeneration presence and density in the UK

Allbwn ymchwil: Cyfraniad at gyfnodolynErthygladolygiad gan gymheiriaid

StandardStandard

Methods for predicting Sitka spruce natural regeneration presence and density in the UK. / Bianchi, Simone; Hale, Sophie; Gibbons, James.
Yn: iForest, Cyfrol 12, 05.2019, t. 279-288.

Allbwn ymchwil: Cyfraniad at gyfnodolynErthygladolygiad gan gymheiriaid

HarvardHarvard

APA

CBE

MLA

VancouverVancouver

Bianchi S, Hale S, Gibbons J. Methods for predicting Sitka spruce natural regeneration presence and density in the UK. iForest. 2019 Mai;12:279-288. Epub 2019 Mai 23.

Author

Bianchi, Simone ; Hale, Sophie ; Gibbons, James. / Methods for predicting Sitka spruce natural regeneration presence and density in the UK. Yn: iForest. 2019 ; Cyfrol 12. tt. 279-288.

RIS

TY - JOUR

T1 - Methods for predicting Sitka spruce natural regeneration presence and density in the UK

AU - Bianchi, Simone

AU - Hale, Sophie

AU - Gibbons, James

PY - 2019/5

Y1 - 2019/5

N2 - Natural regeneration is crucial for silvicultural approaches based on the continuous presence of a forest cover, or Continuous Cover Forestry (CCF). Sitka spruce (Picea sitchensis), is the main commercial species in the United Kingdom (UK), and its potential for CCF has been demonstrated in various studies. However, there are no quantitative models available to predict its natural regeneration in the country. We describe models for Sitka spruce seedlings presence and density under canopy cover in the UK forests, to be used as a substitution of a regeneration survey. Using a natural regeneration dataset comprised of 340 plots, a Generalized Linear Mixed Model (GLMM) was calibrated to estimate the likelihood of regeneration presence at plot level. Seedling density was simulated in a subsequent step using only the subset of data with regeneration presence (138 plots): we compared methods based on GLMMs calibrated to the observed seedling density, and the simple generation of random numbers similar in distribution to the observed values. We validated the models with a cross-validation method using the calibration dataset, and with an independent dataset of 78 plots collected in forests already in the process of transformation to CCF. The best GLMM for regeneration presence included age of the plantation, time after last thinning, favourable ground cover and basal area. After the cross-validation, 73% of the plots were correctly estimated (76% for presence of regeneration and 71% for the absence). After the independent validation process, 82% of the plots were correctly estimated, although 100% for presence of regeneration and only 12% for the absence. Both methods for estimating seedling density had a poor performance, both with the cross-validation and independent validation.The results showed that the tools here described are appropriate for estimating regeneration presence in traditional Sitka spruce plantations. However, alternative methods are required for forests already in an advanced stage of transformation to CCF systems.

AB - Natural regeneration is crucial for silvicultural approaches based on the continuous presence of a forest cover, or Continuous Cover Forestry (CCF). Sitka spruce (Picea sitchensis), is the main commercial species in the United Kingdom (UK), and its potential for CCF has been demonstrated in various studies. However, there are no quantitative models available to predict its natural regeneration in the country. We describe models for Sitka spruce seedlings presence and density under canopy cover in the UK forests, to be used as a substitution of a regeneration survey. Using a natural regeneration dataset comprised of 340 plots, a Generalized Linear Mixed Model (GLMM) was calibrated to estimate the likelihood of regeneration presence at plot level. Seedling density was simulated in a subsequent step using only the subset of data with regeneration presence (138 plots): we compared methods based on GLMMs calibrated to the observed seedling density, and the simple generation of random numbers similar in distribution to the observed values. We validated the models with a cross-validation method using the calibration dataset, and with an independent dataset of 78 plots collected in forests already in the process of transformation to CCF. The best GLMM for regeneration presence included age of the plantation, time after last thinning, favourable ground cover and basal area. After the cross-validation, 73% of the plots were correctly estimated (76% for presence of regeneration and 71% for the absence). After the independent validation process, 82% of the plots were correctly estimated, although 100% for presence of regeneration and only 12% for the absence. Both methods for estimating seedling density had a poor performance, both with the cross-validation and independent validation.The results showed that the tools here described are appropriate for estimating regeneration presence in traditional Sitka spruce plantations. However, alternative methods are required for forests already in an advanced stage of transformation to CCF systems.

M3 - Article

VL - 12

SP - 279

EP - 288

JO - iForest

JF - iForest

SN - 1971-7458

ER -