Recipient Organization
UNIVERSITY OF WASHINGTON
4333 BROOKLYN AVE NE
SEATTLE,WA 98195
Performing Department
Forest Soils
Non Technical Summary
The City of Sammamish is currently creating an Urban Forestry Master Plan. The City faces a number of challenges -- including a pressing public desire for tree preservation in face of development pressure and a widespread fungal pathogen called laminated root rot -- as well as opportunities -- including upcoming land acquisitions.This research will inform and support the City's urban forestry planning and management efforts with best available science. The first part of the project includes 1. creating a baseline map of current canopy cover using high resolution satellite-based imagery, and 2. providing scientific guidance to the City during their update of current urban forestry policies. The second part of this project will address scientific uncertainty surrounding fungal pathogens in urban forests, particularly laminated root rot (Phellinus genus). Infected trees have a very high mortality rate and are often knocked down in high winds, endangering property and lives. Unfortunately while it is critical to find infected trees early, arborists can only detect the pathogen in large trees a decade or more after the tree is infected. A method is needed that can reliably determine the health status of coniferous trees (particularly Douglas Fir), which can be used to verify the traditional diagnosis methods used by arborists. Our proposed research will evaluate the traditional diagnosis methods used by arborists and develop a set of methods that can reliably determine the infection status of specific coniferous trees and accurately map the extent of infected patches using DNA based tools.
Animal Health Component
100%
Research Effort Categories
Basic
(N/A)
Applied
100%
Developmental
(N/A)
Goals / Objectives
The goal is to inform and support the City of Sammamish's urban forestry management planning efforts to safeguard the long-term preservation of the City's forested character.
Project Methods
Canopy and Land Cover MappingMethods• Acquire 1m resolution, four band data or better satellite imagery, along with related products at other scales (several periods of NDVI data, etc)• Develop a comprehensive and exclusive set of land cover classes, in cooperation with the City of Sammamish (City).• Perform an initial training and analysis of the canopy cover based on provided satellite imagery, resulting in a multiclass cover dataset that includes, at a minimum, canopy cover and impervious surfaces.o Software: QGIS (open-source), Orfeo ToolBox (open-source), Google Earth Engine• Perform an accuracy assessment based on the results of the initial training and analysis. The sample evaluated using accuracy assessment will be determined based on a stratified random sampling of the different land cover classes.• If desired by the City, and subject to data availability, repeat analysis on an older time step (possible years include 2011, 2013) to enable change analysis.• Using known patches of laminated root rot as a training sample, attempt to develop a method for mapping of stressed Douglas-fir in the source imagery.Policy AnalysisMethods• To provide valuable feedback concerning the impact of fungal pathogens and other ecological concerns on the City's urban forestry management planning, we will find articles, conference papers, and government or consultant reports in the ecological and policy literature. We will also draw on the long experience of the PI, Robert Edmonds, a fungal pathologist of over 30 years.• Key information from these sources applicable to the City's urban forestry management planning will be summarized in a white paper to provide scientifically grounded feedback on current and future urban forestry policies.• We will use this knowledge to advise the City on the Urban Forest Management Plan process and help them integrate appropriate information and methods into the UFMP, related public education and outreach, and policy guidelines.Laminated Root Rot StudyKey Questions1. How accurate are arborist's diagnoses? That is:a. Given that an arborist has classified a tree as likely infected or in a root rot pocket, what is the probability that it is infected or that it is not infected?b. Given that an arborist has classified a tree as near a root rot pocket, what is the probability that it is infected or that it is not infected?c. Given that an arborist has NOT classified a tree as likely infected or in a root rot pocket, what is the probability that it is infected or that it is not infected?2. What is the most economical and accurate way to sample potentially diseased trees?a. Can soil samples be used or are tree tissue samples needed?b. Can root collar samples be used or are root samples needed?3. Does the microbiome (bacteria + fungi) in root tissue differ between trees with and without detectible root rot? In the adjacent soil?4. Does a relationship exist between the microbiomes of the root tissue and adjacent soil?Sampling Design• Our sampling will be confined to one species of tree, Douglas fir (Pseudotsuga menziesii), which is both locally abundant and highly susceptible to LRR.• We will survey trees in two parks (Pine Lake Park and Beaver Lake Park) where arborists have already completed a visual inspection of a subset of the parks' trees.• Based on the arborist's report, we will classify all inspected Douglas fir trees into three categories: probably infected, possibly infected, and uninfected. Specifically, we will rely on the arborist's comments and descriptions of the defects and likelihood of failure, as arborists rarely report explicitly or consistently on root rot infections.• We will use a balanced sampling design to sample trees from each of the three categories from each park. We will attempt to sample 10 trees from each category for a total of 60 trees. Some categories -- particularly, probably, and possibly infected -- will likely be limited by the number of trees available to be sampled. For categories where the sampling frame is larger than the desired sample, we will use random sampling to determine which trees to sample.• For each tree in the study, 7 tissue and 7 soil samples will be collected, for a total of 14 paired samples.o Tissue samples will be taken from three woody roots of each tree, at a distance of 3m and 6m for a total of 6 root tissue samples (Questions 1 thru 4).o We will take one soil sample adjacent to each of the 6 root tissue samples (Questions 2a, 3, & 4).o One composite sample of the root collar tissue will be taken from each tree, each comprising four subsamples taken using a consistent schema (Questions 1 & 2).o One composite sample of the soil immediately adjacent to the root collar tissue samples.Methods• Collection methodso We will collect all samples in an abbreviated time period during clement weather to limit confounding seasonal and weather effects. We will use a composite borer to collect tree tissue samples and a soil probe fitted with a soil sampling sleeve to collect soil samples.o Between samples, we will wash all tools with a bleach solution and/or denatured alcohol. Prior to sampling tree tissue, we will also wash the exterior of the root or trunk with denatured alcohol to reduce the chance of contaminating the tissue sample with organisms living on the exterior surfaces.o Within 1 hour of collection, samples will be placed in a portable cooler containing dry ice to reduce the likelihood of changes in community composition. Samples will be held below -20 degrees Celsius.• Processing methodso Once all samples are collected, they will be batch processed such that each sample remains at room temperature for an equal and limited amount of time.o Once thawed, soil samples will be homogenized and divided in half. One half of each soil sample will be air dried to prepare for CNH and pH testing; the other half will be wet sieve processed for rRNA extraction. CNH analysis will be conducted using dry combustion, while pH will be measured using a HM Digital model PH-80 pH meter.o Tree tissue samples will be processed using cryogenic grinding, whereby liquid nitrogen is used to freeze tissue samples and a mortar and pestle are used to grind the samples (see e.g. http://opsdiagnostics.com/notes/leaftissue.htm). Although other more sophisticated methods have been suggested in the literature (e.g. Lorenz et al 2010: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3152223/), they are not feasible here.o The wet sieved soil samples and ground tree tissue samples will be sent to Noah Fierer's lab at University of Colorado, Boulder for RNA extraction and analysis using an Illuminata MiSeq next generation system.o Fungal and bacterial community data will be derived by first demultiplexing the raw data using a custom Python script (available at: https://github.com/leffj/helper-code-for-uparse; Edgar 2013) and then using the R based DADA2 pipeline to transform demultiplexed data into a community data table. Fungal taxonomic information comes from the UNITE database (available online at https://unite.ut.ee/), while bacterial taxonomic information comes from the GenBank database.• Statistical methodso The datasets will not be rarefied following recent statistical developments in the literature (McMurdie and Holmes 2014, Hart et al 2015).o We will use Bayesian updating to estimate the relevant parameters for Question 1 above.o We will use ANOVA or an ANOVA variant (e.g. PERMANOVA) depending on the error structure and distribution of the data to address Question 2.o We will use multivariate community ecology tools, including cluster analysis, NMDS, indicator species analysis, PERMANOVA, and differential expression analysis to address Questions 3 & 4 above.