Recipient Organization
UNIVERSITY OF VERMONT
(N/A)
BURLINGTON,VT 05405
Performing Department
School Of Natural Resources
Non Technical Summary
Soils are the largest terrestrial carbon (C) sink, thus even small changes in soil C canaffect terrestrial C feedbacks to the atmosphere (Jobaggy and Jackson 2000,Kirschbaum 2006). While forest regrowth following agricultural abandonment in theNortheast USA has been a major C sink over the last century (Myneni et al. 2001;Goodale et al. 2002), there has been less focus on northern forest soils.Thus, animportant question remains, are northeastern forest soils a C sink and what willthey be in the future?Documenting how afforestation and forest management affect soil C, a criticalforest service, across scales is a research priority locally and globally (see lettersof support; Jandl et al. 2007; Edenhofer et al. 2014). C storage in forest soils farexceeds C storage in aboveground biomass (Adair et al. 2018). Furthermore, variablesthat predict soil C at small scales (e.g., stand level) often differ from those that predict atlarger scales (e.g., regionally). Enhancing forest soil C carries many benefits,including climate change mitigation via CO2 sequestration (Jandl et al. 2007, Lange etal. 2015), the promotion of forest health and productivity via increasing soil quality(Melillo et al. 2002), and the regulation of water flow and nutrient fluxes (Aber et al.1995, Lal 2016). For these reasons, public and private land management agencieswould like to include enhancing C storage as part of their management strategy. Thus,our proposed work is timely.Forest soil C is an ecosystem property. As such, it is controlled by a combination ofdrivers, known as 'state factors' (Vitousek 2004; Chapin et al. 2011). In the state-factorframework, climate, organisms, topography, soil geological substrate, and time sincedisturbance are independent factors controlling soil C. Traditionally, ecosystemecologists have focused on biotic and abiotic drivers of process separately − with bioticdrivers (e.g., organisms) driving local patterns and abiotic drivers (e.g., climate) drivingregional patterns. However, recent work suggests that the relative importance ofthese drivers − including management and properties such as soil type −changes from local to regional to global scales (Hooper et al. 2012; Adair et al.2018; Thompson et al. 2018). We propose to determine predictors ofsoil C storage to enable managers to promote soil C accrual at stand and regionalscales in Vermont and NE forests. While doing this, we will explore why drivers of soil Cvary with scale.Our proposed work falls under the research priorities of sustainable forest management,forest health and productivity, and forest ecosystem services, aligning with RSENR'sstrategic plan in key ways: (1) we will use Vermont and RSENR forests to giveundergraduate, graduate, and postdoctoral participants hands-on experience with fieldandlab-based teamwork, quantitative methods, science communication, andcollaboration across disciplines; (2) our proposed work advances integrated researchaddressing environmental and natural resources challenges faced by local and regionalcommunities; (3) this project will establish a new cross-disciplinary collaboration amongfive PIs (three women; two early career). We will leverage our diverse skill sets inforestry, ecosystem ecology, remote sensing, biodiversity, and modeling to grow ourlocal and national reputation for impactful forestry research and outreach in Vermont.
Animal Health Component
50%
Research Effort Categories
Basic
10%
Applied
50%
Developmental
40%
Goals / Objectives
Objectives and Overarching Goal:Identify how drivers of forest soil C vary across local to regional scales to predict andpromote soil C in managed forests in Vermont and the northeast region.We have three core questions:Q1) How do forest management and edaphic factors directly and interactively influenceC storage in regrowing forests?Q2) How do patterns and drivers of forest soil C vary from local to regional scales?Q3) What are the management implications from improved understanding of soil C?We hypothesize that:H1) Characteristics of forest management (e.g., ownership, stand age) and biophysicalfactors (e.g., elevation, aspect, soil texture) will play important roles in determiningwhether forest soils are C sinks (Q1).H2) Patterns and drivers of forest soil C will vary with scale, with climate andmanagement controlling regional patterns and biophysical variations (e.g.,biodiversity, forest composition, topography) within sites controlling local-scalepatterns (Q2).H3) Projected changes in climate and management will have long-term implications onforest soil C accumulation and storage across the NE (Q3).We have identified the following research objectives to address our questions andhypotheses:Objective 1 (Q1, H1): In collaboration with our cooperators andregional stakeholders, we will establish a comprehensive database of soil C (iSoC) at stand, local, and regional scales. Specifically, we will leverageexisting and new data to: (1) gather intensive stand-level data in dominant forest typesacross a range of stand ages, with a focus on UVM Research Forests; (2) create anextensive NE soil C survey building on the work of Adair, who collected data at 34 sitesacross Vermont and has obtained NE forest soil C data from the USDA ForestInventory and Analysis (FIA).Objective 2 (Q2, H2): We will use the multi-scalar datasets collected in Objective 1 toexplore how drivers (e.g., management, forest type) of soil C change with scale (fromstand to forest to region). We expect important controlling factors to include climate,soils (% clay, pH, organic matter, aggregate formation), topography (relief andelevation), management, and forest age.Objective 3 (Q3, H3): We will use the multi-scalar drivers defined in Objective 2 incombination with climate and management scenarios eveloped with our cooperators tounderstand and predict forest soil C using statistical and process-based models.Favored forest ecosystem models in this region include the LANDIS-II model, which hasbeen used successfully to predict forest dynamics (Duveneck et al. 2017), but has notbeen parameterized for forestsoils. Modeling will allow us toestimate current stores of soil Cand investigate changes in soil Cunder various climate andmanagement scenarios.
Project Methods
Integrated Soil Carbon (iSoC) inventory:The Integrated Soil Carbon (iSoC) inventory will be the most comprehensive collection of soil C data for northeastern Forests. The iSoC will represent stand level data fordominant NE forest types using UVM Research Forests and Natural Areas, FEMC sites,USDA FIA plots, and LTER sites. The inventory builds on Adair's Vermont soil C surveyand will incorporate FIA soil data (obtained by Adair). The minimum data forinclusion are climate, topography, pH, stand age, and soil texturewith additionalinformation collected as available.Intensive stand level data:These data will be collected in dominant NE forest types, including northern hardwood,mixed-wood, spruce-fir, and oak-pine-hemlock, using UVM Research Forests, NaturalAreas, and FEMC monitoring plots whenever possible. In each forest type,sites will beselected along a forest age gradient, to address the critical question of whetherregrowing forests are C sinks. Sites will be selected based on the data gathered duringour extensive forest survey (see below). At these sites, we will measure soil C in thewhole soil as well as in aggregate fractions and characterize C storage in woody debris.To better understand the stand-scale controls on soil C, we will intensively measure asuite of biotic and abiotic factors hypothesized to increase or decrease soil C at thesesites. Biotic measurements will include metrics of plant and microbialcommunity abundance, diversity, and activity; abiotic measurements will include soilcharacteristics (e.g., mineralogy, pH, and fertility, aggregation), topography, land usehistory, and microclimate.Extensive forest soil data:We will leverage soils data gathered by Adair and existing FIA soils data, byadding a variety of sites in uncharacterized and under-sampled areas. Here, in additionto soil and woody debris C, we will focus on gathering easy-to-obtain data that will beused to predict soil C at larger scales such as climate, stand age, topography, soil pH,understory plant diversity, and soil texture. We will work with FEMC and otherorganizations to co-locate soil sampling with sites that have been characterized forforest composition and stand age. These data will be used to investigate how drivers ofsoil C in NE forests change across stand, forest, to regional scales.Statistical Methods: Drivers of Soil Carbon SequestrationDrivers of soil C are most likely scale- and site-specific (Huang et al. 2017). Thus, ifanalyzed only at the largest scale, positive and negative correlations at different scalesmay cancel each other out, showing weak or no relationships (Biswas and Si 2011). Wewill use statistical methods to understand the factors ("drivers") that control soil Cstorage at local to regional scales. We will run statistical analyses using BayesianWeights of Evidence for spatial probability analyses (e.g., Galford et al. 2015, Sonter etal. 2017) and a combination of model selection and structural equation modeling (SEM)methods (e.g., Adair et al. 2018). Ancillary site data will be pulled from existingdatabases or populated from spatial data (e.g., elevation from DEM if not reported). Wewill test the drivers in the iSoC inventory for spatial autocorrelation using Crammer'sCoefficient (Galford et al. 2015), removing any drivers with high (<0.3) spatialautocorrelation. Each remaining driver will be used in the statistical analyses, wherethey will be tested for significance at various scales of analysis. We will investigate howstatistical approaches yield similar or different results depending on scale, which willreveal the scales at which each driver is important.Additionally, we will investigate empirical methods of defining the scales at which different drivers operate (e.g., empirical mode decomposition, EMD, as in Huang et al.2017; EMD uses the data to tell us the important scales for soil C and its covariates; it islike a Fourier transform but with fewer assumptions that need to be met). The result ofour analyses will be a strong understanding of the relative importance of the factors thatinfluence forest soil C storage at different scales.Modeling: Management ImplicationsModeling will allow us to estimate current stores of soil C and investigate changes in soilC under climate and management scenarios through the year 2060. Multi-scalar dataand analyses will allow us to make recommendations for managers attempting topromote soil C storage from stand to regional scales.We will validate and improve ecosystem models of forest C dynamics. Statisticalrelationships among soil C and drivers (see above) will be used to validate previousregional modeling simulations, which currently do not have strong parameterization forsoil C. Model parameterizations will be improved as needed. Our team has experienceconducting such work with a wide range of ecosystem models including theDeComposition DeNitrification model (DNDC; Costa et al. In Review), DAYCENT (Chenet al. 2016), Terrestrial Ecosystem Model (TEM, Galford et al. 2011), CORPSE,MIMICS, and MEND (Moore et al. 2015; Moore et al. in review; Sulman et al. 2018), andLANDIS-II with PNET (Duveneck et al. 2017).Forest cover scenarios for the region, provided by Collaborator Thompson (Duveneck etal. 2017), have been used as inputs to LANDIS but without parameterization for thesoils module. We will provide parameterization for soil C and will use regionaldownscaled climate scenarios to examine interactive effects of management andclimate on future C storage. Climate drivers will come from the EPA's EnviroAtlasClimate Scenarios (EPA 2013) generated from the NASA Earth Exchange DownscaledClimate Projections ensemble averages (30 models) for various representativeconcentration pathways (RCPs). This datais publically available at 30 arc-seconds(~800 m2), which is sufficient for this type of soil C modeling (Thrasher and Nemani2013).Open Access and Data Archive:Project data will be archived with the Forest Ecosystem Monitoring Cooperative (FEMC,https://www.uvm.edu/femc/). The FEMC data archive provides searchable and linkedinformation for research projects, including document archive, data display andvisualization, search and download features, and metadata summaries. The websiteallows users to browse and search, and is indexed by search engines such as Google.The FEMC follows the Ecological Metadata Language standard for metadata, which isthe preferred standard for the Long-Term Ecological Research Network. Data will beavailable for exploration and download, following the Creative Commons(https://creativecommons.org/) licensing scheme.Plan for inclusion of stakeholder input:This work will build on existing collaborations among the PIs and various stakeholdergroups including FEMC, Harvard Forest, the Northern Institute of Applied ClimateScience, and the USDA Northern Forests Climate Hub (see support letters). We willcollaborate closely with our stakeholders to identify research sites, relevantmanagement scenarios, and other issues of concern. We will also work with them todisseminate results in a useful form.