Date of Award

Summer 2019

Project Type


Program or Major

Natural Resources and Environmental Studies

Degree Name

Doctor of Philosophy

First Advisor

Mark Ducey

Second Advisor

Scott Bailey

Third Advisor

Thomas Lee


Forest land managers need ecological classification to assess and describe resource conditions, vegetation conditions, outcomes resulting from various management prescription scenarios, and communicate environmental effects of land management planning alternatives. However, there is a need to incorporate more ecological classification into the land management plans. The U.S. Forest Service’s approach, the Terrestrial Ecological Unit Inventory (TEUI), relies heavily on field data collection and verification of map unit delineations that is time-consuming and costly. Traditional mapping methods far exceed the current financial capacity of the U.S. Forest Service. In order to justify new ecological classification mapping approaches, there needs to be significant evidence that new approaches will create equivalent or superior map products, reduce costs, improve efficiencies and maybe improve accuracy. Therefore the objectives of chapter 2 were to use the Soil Inference Engine (SIE) to partition the areal extent of a project area watershed in the White Mountain National Forest (WMNF) using on topographic metrics derived from Light Detection and Ranging (LiDAR) data including both timber managed and un-managed timber production lands. A total of 189 plots were randomly generated within strata, based on parent material, and topographic metrics using a stratified random sampling approach. The number of plots calculated for stratified random sampling was predominately determined by the number of strata, the acres of timber-managed areas, and budget. 172 of those plots had both vegetation and soils information recorded. The results from chapter 2 showed that stratified random sampling using LiDAR-derived topographic metrics as SIE data inputs were sufficient in capturing the environmental gradients required by the U.S. Forest Service ecological classification requirements. Additionally, 10 New Hampshire Natural sensitive indicator species were located and recorded in 16% of plots stratified by topographic metrics and parent material. These results suggest this new approach to ecological classification on the WMNF improved the accuracy and efficiency in delineating ecological areas as well as locating the presence of nutrient rich areas.

The objectives of chapter 3 used nonmetric multidimensional scaling (NMDS) to assess the relationship between understory species and environmental variables, including parent material, slope, aspect, elevation, and wetness. The results from chapter 3 showed how both soil properties and topographic metrics correlated with understory species in ordination space. NMDS ordination explained 81.1% of the cumulative variation of understory species presence in three dimensions using soil properties and topographic metrics with a final stress value of 17.3 and a p-value of 0.04. NMDS results also suggested that understory species clustered distinctly within New Hampshire Natural Community types. These results also support the idea that LiDAR-derived topographic metrics could assist in determining where community types are positioned across a landscape. Additional NMDS analysis also showed either soil chemistry or topographic metrics explained nearly equal amounts of cumulative understory species variation. The results from this objective highlights the use of topographic metrics as predictors of understory vegetation, and likely community types, which could be validated in other WMNF watersheds.

Finally, the primary challenge for ecological classification is reducing the cost of traditional unit mapping. Therefore, the objectives of chapter 4 was a conceptual synthesis of the reasoning behind doing ecological classification. Information from the WMNF management plans of 1985 and 2005, and current National and Regional land management direction of the US Forest Service were reviewed. A cost review of ecological classification by stratified random sampling using LiDAR-derived topographic metrics was compared to traditional TEUI mapping methods. In both approaches, the mapping of the plots averaged approximately $989.00 per plot including soil chemistry analysis from U.S. Forest Service Laboratory.

This yielded a total cost of approximately $623,000 for the traditional TEUI compared to approximately $221,000 including the LiDAR acquisition required for stratified random sampling using topographic metrics. This chapter showed that stratified random sampling using LiDAR-derived topographic metrics reduced costs by approximately $402,000, including the additional LiDAR acquisition costs, compared to the traditional TEUI approach.