Using a Species Distribution Model and Site-Specific Microclimatic Variables to Model Presence of Spodic Soil Properties and Relative Occurrence Rate of Picea rubens (Red Spruce) to Inform Red Spruce Management
Date of Graduation
Davis College of Agriculture, Natural Resources and Design
James A Thompson
Pamela J Edwards
Michael P Strager
Red spruce restoration efforts in the central Appalachians are of interest to land managers because of the reduced current extent and the numerous ecosystem services provided by red spruce forests. These land managers require information regarding the best places to focus restoration efforts. Studies attempting to locate optimal locations for red spruce restoration have been conducted to in the central Appalachians, most of which utilize modeling. In particular, podzolization, a soil formation pathway present under conifer vegetation in the central Appalachians has been used to help select areas for red spruce restoration. The effectiveness of using recent podzolization (as evidenced by spodic soil properties) to predict historic vegetative cover has been found to be useful in prioritizing areas for red spruce restoration. The objectives of this research were twofold: evaluate the efficacy of the model, Maximum Entropy (MaxEnt), to model presence of spodic soil properties and evaluate the efficacy of modeling red spruce relative occurrence rates using topographic and microclimatic variables.;For the first objective, MaxEnt was used to model presence of spodic properties in 124,687 ha in the central Appalachians using 221 presence-only soil observations and a suite of topographic and satellite-derived variables. Results of this study were compared to a model output generated using Random Forest (a presence/absence model). The results showed approximately 62% agreement (both models predicted high, or both predicted low probability of presence), and 38% disagreement (one model predicted high probability of presence, while another predicted low, or vice-versa). However, without field validation, it is not known which model output is better. No variables used in this exercise were found to be particularly important to spodic property presence, which is likely due to the relatively coarse scale of variables used.;To evaluate the second objective, air temperature, soil temperature, and soil moisture were measured in situ for approximately one year in a small (5.4 km2) high elevation (700--900 m) watershed in the central Appalachians. The raw data collected was summarized into variables believed to be important to red spruce relative occurrence rates and largely based on available literature. Four preliminary MaxEnt models were run using (i) only topographic variables, (ii) only air temperature variables, (iii) only soil temperature variables, and (iv) only soil moisture variables. The most important variables (as evidenced by permutation importance value) were utilized in a final model run. Altitude above channel network was the most important variable in the preliminary run which utilized only topographic variables, and the final run which used the most important variables. The relationship between altitude above channel network and red spruce relative occurrence rate was inverse: as altitude above channel network increased, red spruce relative occurrence rate decreased. The second most important in the final model run was August absolute maximum air temperature. Lack of importance of other microclimatic data is most likely due to poorly interpolated surfaces and missing data. Model outputs from both the preliminary run that utilized only topographic variables, and the final run both predicted the lowest red spruce relative occurrence rate at the highest elevations in the watershed on ridgetops and shoulders. The highest red spruce relative occurrence rates were found at the lowest elevations of the watershed. This, coupled with the importance of altitude above channel network, suggest that higher red spruce relative occurrence rates occur in concave landscape positions that promote cooler air and soil temperatures, and increased soil moisture. Agreement between the preliminary model run that used only topographic variables and the final model run was approximately 82%, while disagreement between the two was only 18%. The limited success of creating soil temperature and soil moisture variables, coupled with the fact that there was little difference between the model that utilized only topographic variables and the model that incorporated microclimatic variables, suggests that it may be feasible to utilize only topographic variables in future efforts. If microclimatic variables are desired, air temperature was found to be important in this model, and would be easier to measure in the field. Red spruce restoration should continue to target the highest elevations of the central Appalachian landscape, but should not necessarily be limited to the highest ridgetops and shoulder landscape positions. Instead, red spruce restoration should target high elevation concave landscape positions like cold air drainage ways which promote cooler air and soil temperatures as well as soil moisture.
Nottingham, Adrienne, "Using a Species Distribution Model and Site-Specific Microclimatic Variables to Model Presence of Spodic Soil Properties and Relative Occurrence Rate of Picea rubens (Red Spruce) to Inform Red Spruce Management" (2017). Graduate Theses, Dissertations, and Problem Reports. 6329.