Oak recruitment in young forests: modeling the role of forest characteristics before disturbance and their interactions with the abiotic environment

Restricted (Penn State Only)
- Author:
- Gershey, Joshua
- Graduate Program:
- Ecology
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- August 22, 2024
- Committee Members:
- Laura P Leites, Thesis Advisor/Co-Advisor
Duane R Diefenbach, Committee Member
Jason Kaye, Program Head/Chair
Kimberly Bohn, Committee Member - Keywords:
- ecology
forest ecology
oak
oak recruitment
mixed oak
random forest
environment
oak regeneration
silviculture
forest management
forestry
stem exclusion
classification
forest stand
quercus
pennsylvania
PA
SILVAH
random forests
RF - Abstract:
- Natural forest stand development is a complex process comprised of multiple stages throughout a forest’s lifecycle where biotic and abiotic factors can shape forest composition and canopy strata. This study focuses on oak (Quercus spp.) recruitment, 10 to 15 years after overstory removal, in managed Pennsylvania mixed-oak forests. This study uses biotic data collected before and after overstory removal, in 96 mixed-oak Pennsylvania state forest stands and abiotic data from publicly available climate, soils, and landscape datasets. In Chapter 1, I describe the complex process of forest stand development from stand maturity to stem exclusion, as it relates to key biotic and abiotic factors, and summarize literature to assess the current state of oak regeneration and recruitment models. This literature review identified a need for comprehensive recruitment models that incorporate explicit climate and soil variables and use stand as the modeling unit. In Chapter 2, I address this need and modeled oak recruitment 10 to 15 years after overstory removal as a function of biotic and abiotic factors and explore how they contribute to the recruitment of oaks in the dominant/codominant canopy of the young forest. I present two recruitment classification models, one for mixed-oak and one for chestnut oak (Q. prinus), developed using Random Forests and a suite of biotic and abiotic predictors. The models classified stands as low presence (≤ 20% dominant/codominant canopy mixed-oak or chestnut oak) or high presence (> 20% dominant/codominant canopy mixed-oak or chestnut oak). Individual variable contribution was assessed through partial dependency plots, which show how the likelihood of a stand being classified as high or low presence changes across the range of each variable. Potential variable interactions were assessed through bivariate dependency plots, showing the change in classification across the range of two predictor variables. A conceptual diagram was developed to outline key ecological factors affecting oak recruitment. Partial dependency plots and the two models show how biotic and abiotic factors contribute to stand composition at approximately the stem exclusion stage. In Chapter 3, I describe typical values characterizing stands with two levels of oak recruitment and highlight key differences in stands predicted to have a high presence of dominant/codominant oaks from those predicted to have a low presence of dominant/codominant oaks. These typical values may provide insight for forest management organizations tasked with landscape-level forest planning and reinforce the reliability of decision-support tools.