22 January 2013 Science Briefs

SNAMP PUB #16: Characterizing habitat associated with fisher den structures in the Southern Sierra Nevada, California using discrete return lidar

Article Title: Characterizing habitat associated with fisher den structures in the Southern Sierra Nevada, California using discrete return lidar.

Authors: Feng Zhao, Rick A. Sweitzer, Qinghua Guo, Maggi Kelly.

Research Highlights:


  • Discrete return lidar-derived variables proved useful for characterizing the forest habitat conditions surrounding trees and snags used by the Pacific fisher.

  • Large trees and snags used by fishers as denning structures were associated with forested areas with relatively high canopy cover, large trees, and high levels of vertical structural diversity. Den structures were also located on steeper slopes, potentially associated with drainages with streams or access to water.

  • We hypothesize that the types of structurally complex forest habitats occupied by denning female fishers provided access to an abundance and diversity of prey required for successful reproduction.


Background:
Fishers (Martes pennanti) are a medium-size mammal carnivore with a greatly reduced range in Sierra Nevada forests. Because of continued concern over the limited abundance and range extent of this species in western North America, several major studies are underway aimed at identifying population limiting factors, general habitat associations, and specific habitat requirements for resting and reproduction. Resting and denning structures are considered to be the most important habitat elements required for maintenance of fisher populations. In this paper, we explored the utility of small footprint discrete return lidar (light detection and ranging) remote sensing data to assess the denning habitat of fishers in the Sierra National Forest. We examined the ability of lidar-derived metrics to capture topography and forest structure surrounding denning trees used by the Pacific fisher as a case study to illustrate the utility of lidar remote sensing in studying mammal-habitat associations.

Objectives


  1. Identify lidar-derived variables that separate fisher denning structure from random trees; and,

  2. Test the utility of lidar data to assess fisher denning habitat structure across scales from immediately surrounding a tree to 50 meters away from it.

Methods
We mapped 28 natal and maternal den trees in the Sierra Nevada Adaptive Management Project (SNAMP) Sugar Pine area used by female fisher in 2008, 2009, 2010, or 2011. These are our “positive” cases. Locations of trees or snags that were confirmed as den trees were recorded using a Garmin 60CSs handheld GPS. We also mapped numerous trees that were not used by fishers. From these, we randomly selected an equivalent number of trees – these are our “negative” cases.

To explore the difference between these trees, we used Classification and Regression Trees (CART, a statistical classification model that compares positives and negatives) to statistically compare the slope and lidar-derived forest height and structure metrics in the circular area (with radius of 10–50 meters) surrounding denning trees and randomly selected non-denning trees. We tested our model accuracy using a range of standard validation methods, including substitution and cross-validation.

Results:
Our results show that there is a strong association between fisher denning activity and the surrounding forested environment across scales (from 10 – 50m surrounding a tree). We achieved high classification accuracy (overall accuracies above 80%) at 20, 30 and 50 m ranges. The best classification accuracies were found at 20 m surrounding trees. Tree height and slope were important variables in classifying the area immediately surrounding denning trees; at scales larger than 20 m, forest structure and forest complexity became more important factors.

The following variables were useful in predicting the difference between presence trees and absence trees. Some of these variables are common in forestry, such as canopy cover and slope; some are not as common as they refer to the pattern of lidar returns. These variables, such as transmittance, and standard deviation of the lidar returns would have to be recreated with lidar data.


  • 10 meters surrounding a tree: tall trees (over 33 meters) and slope greater than 17 degrees;

  • 20 meters surrounding a tree: the standard deviation of the lidar returns and slope greater than 16 degrees;

  • 30 meters surrounding a tree: the standard deviation of the lidar returns, the distribution of heights of all trees in the plot, and slope greater than 13 degrees;

  • 40 meters surrounding a tree: the standard deviation of the lidar returns and slope greater than 17 degrees;

  • 50 meters surrounding a tree: the mean height of trees in the plot greater than 20m, the transmittance of lidar, and canopy cover greater than 0.99.

Conclusions:


  1. Discrete return lidar-derived variables proved useful for characterizing the forest habitat conditions surrounding trees and snags used by the Pacific fisher.

  2. Large trees and snags used by fishers as denning structures were associated with forested areas with relatively high canopy cover, large trees, and high levels of vertical structural diversity. Den structures were also located on steeper slopes, potentially associated with drainages with streams or access to water.

  3. We hypothesize that the types of structurally complex forest habitats occupied by denning female fishers provided access to an abundance and diversity of prey required for successful reproduction.

  4. The CART algorithm was useful: it handled a broad range of data, was easy to use, and easy to interpret.
  5. It should be noted that this study was limited in both space and time, and to generalize our findings to larger scale will require further exploration of a larger area over a longer term.

  6. Additionally, many of the variables that were important in classifying presence from absence trees are specific to lidar data, and not, as yet, commonly used in forest science.

Full Reference:
Zhao, F., R.A. Sweitzer, Q. Guo, and M. Kelly. Characterizing habitats associated with fisher den structures in the Southern Sierra Nevada, California using discrete return lidar. Forest Ecology and Management 280: 112–119

The full paper is available here.

For more information about the SNAMP project and the Spatial and Fisher teams, please see the: Spatial Team Website, and the Fisher Team Website.

To learn more about lidar data, check out our lidar FAQs sheet. For more information on lidar data in SNAMP, see the spatial team website: http://snamp.cnr.berkeley.edu/teams/spatial, and our spatial team newsletters that focus on lidar: Vol. 2, No. 3, and Vol. 5, No. 1.

No Comments Yet

Post a New Comment