Vegetation Dynamics

A reorganization of the Arctic plant community may be a significant result of climate change that drives important feedbacks to the atmosphere and to permafrost stability (Epstein et al. 2004, Walker et al. 2006, Sturm 2010). Increasing predominance of shrubs over smaller-statured tundra vegetation may create substantial feedbacks to the ecosystem and the climate system through changes in albedo, energy exchanges (water and heat), snow depth, timing and extent of permafrost thaw, water and nutrient availability, microbial activity, and relative CO₂ uptake and release (McGuire et al. 2006). Changes in the presence of mosses in an ecosystem would have subsequent consequences for soil temperature and permafrost stability because of their important role in insulating the soil (Walker et al. 1994). Furthermore, some mosses support methanotrophs, which oxidize methane (Kip et al. 2010), whereas sedges provide a conduit for methane transport from soil to atmosphere (King et al. 1998). Changes in plant community composition, such as a shift to shrub-dominated systems and a reduction in moss cover, will alter evapotranspiration dynamics from evaporation (physically controlled fluxes) to transpiration (physiologically controlled fluxes), which would be critical information for predicting evapotranspiration dynamics with climate and permafrost change. Additionally, shrub-dominated systems will increase carbon gain and landscape-level water-use efficiency. Hence, the structure and function of the plant community and their responses to a changing environment are central to our analyses of biogeochemical cycling, hydrology, and feedbacks between the tundra and the atmospheric and climate systems.

R. Norby (ORNL) will lead this team effort. Norby is a plant physiological ecologist with interest in nitrogen cycling, plant-soil interactions, and ecosystem responses to climate change.

C. Iversen (ORNL) will be responsible for measurements of N cycling in the active layer.

V. Sloan (ORNL), a postdoctoral researcher, will work with Norby and Iversen to characterize plant community composition, root distribution, and plant nitrogen contents.

J. Childs (ORNL)  will provide field and laboratory technical assistance at ORNL.

A. Rogers (BNL) is responsible for plant physiological measurements, including leaf gas exchange and biochemistry.

N. McDowell (LANL) and J. Cable (UAF) will make measurements of water isotopes in plants and whole-system gas exchange.

C. Xu (LANL) will be responsible for integrating plant and soil measurements into a model of C-N interactions.

D. McGuire (UAF), E. Euskirchen (UAF), and D. Hayes (ORNL) will use the data from this task to improve the representation of plant functional types in models.

Identification and quantification of the key processes linking plant community structure and function to soil moisture and nutrient availability are essential for refining mechanistic-based models of arctic ecosystems and for linking biogeochemical cycling models to vegetation dynamics models in an integrated, coupled land-climate model framework for both regional and global scales. The composition of the plant community can be measured directly at local scales, and changes in community composition in response to climate warming and permafrost degradation can be inferred from manipulative experiments and observations across gradients of permafrost degradation; however, representation of plant community function and dynamics is more challenging at the grid-cell scale. Scaling plant function to the grid-cell scale should be based on observable relationships between plant community composition and geomorphic units. Furthermore, a process-based framework is needed for predicting changes in the plant community and associated function as the climate changes and permafrost degrades.

Vegetation Dynamics Goal: Describe and quantify the mechanisms that drive structural and functional responses of the tundra plant community to changing resource availability, in support of a predictive framework for evaluating GHG and energy feedbacks to climate through vegetation dynamics.

Our approach will rely on the use of PFTs, which group plant species according to common morphological or physiological traits (e.g., broadleaf woody plants, sedges, mosses, and lichens). CLM calculates gross primary productivity (GPP), or gross ecosystem C uptake, using functional relationships describing plant photosynthesis in relation to prevailing temperature, light, and foliar nitrogen concentration. The relationship is parameterized for different PFTs, and the fraction of land area populated by different PFTs is used to generate GPP estimates for a grid cell. New fine-scale measurements and process understanding are needed to parameterize CLM for the tundra and provide appropriate boundary conditions for up-scaling beyond the domain of direct measurement. Additional measurements are needed to provide independent model evaluation. Three specific needs are (1) observations to estimate current PFT distribution across the landscape, (2) data to inform functional relationships describing GPP of arctic PFTs, and (3) process understanding to project how PFT distribution will change with permafrost degradation.

Characterize plant community composition. Documentation of the characteristics of the plant community in relation to polygon features will provide the fundamental framework for estimating plant community composition and function at the grid-cell scale and for refining PFTs for predictive relationships.

• Vegetation survey plots (1 × 1 m) will be established at the center, edge, and trough of three to five replicated polygons in each of three polygon types (low-centered, high-centered, and intermediate). Species composition of these plots will be determined by visual estimation of fractional coverage (Fletcher et al. 2012).
• Leaf Area Index (LAI) of plant communities across the polygon gradients will be measured using an LAI-2200 Plant Canopy Analyzer.
• At the end of the growing season, plants in 0.2 × 0.2 m subplots will be harvested and aboveground biomass and leaf area will be measured by species.

Improve PFT definitions. CLM currently uses only two PFTs (one grass and one shrub type) to represent arctic vegetation, greatly limiting its capability to represent arctic plant functions and feedbacks or to simulate arctic response to a warming climate. The 10 arctic PFTs in the Terrestrial Ecosystem Model (TEM) (Euskirchen et al. 2009) allow a superior basis for hypothesis testing of the relevant vegetation parameters, and we will use TEM as guidance for improving the PFTs in CLM. We will augment the PFT definitions with plant parameters that are needed to improve GPP and albedo calculations in CLM. In particular, we will measure the spectral properties and parameters of photosynthetic biochemistry and leaf physiology that are key CLM inputs and that facilitate up-scaling to the landscape level (e.g., leaf mass per unit area, leaf area index, v_c,max, and tissue N concentration; Thornton and Zimmermann 2007, Xu et al. 2012) for a range of tundra plant species across different PFTs, measured under arctic summer conditions across the gradients created by high-centered and low-centered polygons and other thermokarst features.

• Focusing on key plant species representing different PFTs, use a LI-COR 6400XT gas exchange system to measure CO₂ assimilation in relation to internal leaf CO₂ concentration, from which v_c,max can be calculated. Measurements will be made three times during the growing season.
• Measure N concentration, leaf mass per unit leaf area, and derive the fraction of leaf N invested in Rubisco (fN_Rubisco) in the leaves used in the gas exchange measurements.
• Measure spectral characteristics, including albedo, of individual leaves and mixed-species plant communities using handheld and track-based scanning spectroradiometers throughout the snow-free season. Foliar N concentration of the scanned leaves will then be determined.

Make PFTs dynamic. The primary data needed for estimation of PFT distribution across the landscape are assessments of plant community composition (fractional cover) across the thermokarst gradients in different geomorphic units, as described above. We also need data and process understanding to enable predictions of changes in plant community composition as permafrost degrades in a warming climate. We have developed a working hypothesis that permafrost degradation causes a change in water and N availability and distribution that will drive changes in PFT distribution across the landscape. The data needed to test this hypothesis and to develop the functional relationships for modeling include seasonal variation in active-layer N availability, plant-soil feedbacks that alter C-N cycling and N availability, plant use of available N (including seasonal dynamics, root distribution, and N fixation), and root distribution of plants in relation to available water. Our objective will be to establish a new set of PFTs based on N acquisition and allocation rather than plant morphology. The research will be guided by a plant physiological model of C-N interactions (Xu et al. 2012).

• At peak standing crop at the end of the growing season, soil cores associated with each aboveground community measurement will be used to assess community-level root biomass and rooting depth distribution. These samples will also determine the N concentration and content of belowground biomass throughout the active layer. We will also carefully excavate root systems of the most important species in order to determine species-specific root distribution and N concentration of roots with depth in the active layer. Species-specific rooting characteristics will help us to better understand and project the causes and consequences of changes in belowground biomass and N content in response to permafrost degradation. Root distribution of different species will be analyzed in relation to soil water availability.
• Seasonal variation in plant-available nutrient concentrations in the active layer will be measured using ion-exchange resins (Giblin et al. 1994, Natali et al. 2011) at locations adjacent to vegetation survey plots. The resins, which provide a time-integrated measure of plant-available N, P, and other elements, will be deployed from mid-June to August, August to October, and October to June in order to capture seasonal dynamics.
• Forms of nitrogen available in the active layer will be assessed as extractable concentrations of organic N, NH₄⁺, and NO₃^– in a subsample of soil taken from the cores used to determine rooting biomass and depth distribution. These data will also be analyzed in relation to landscape and plant community characteristics and will be used to inform nutrient cycling rates in models.
• Plant influence on C and N metabolism in the active layer will be measured in soil sampled from plant communities occurring across the sequence of permafrost degradation. Soil cores will be used to sample active layer soils from two different depth intervals, and root-free, homogenized soil will be incubated under standard laboratory conditions (e.g., Iversen et al. 2012). Incubations will be conducted both aerobically and anaerobically to assess the potential influence of saturated soil conditions, as well as across a field-relevant range (i.e., from –2°C to 10°C) of temperatures in order to provide a temperature response surface for model parameterization. CO₂ and CH₄ emission will be measured by gas chromatography, and net NH₄⁺ and NO₃^– mineralization rates will be assessed using an autoanalyzer to determine the difference in KCl-extractable NH₄⁺ or NO₃^– over time as compared with initial samples. Changes in total N over time will also be assessed in order to determine the relative importance of organic compared with inorganic N at a given time. Plant detritus (leaf litter, roots) of different PFTs (shrubs, sedges, moss) will be added to some samples to determine whether increased biomass and litter production of different PFTs will affect soil nitrogen cycling.
• Nitrogen fixation activity in root systems, soil, and bryophytes will be assayed using the acetylene reduction approach (Hardy et al. 1968). Samples will be incubated in a 10% acetylene atmosphere, and ethylene production will be measured by gas chromatography.
• Plant C and N metabolism will be measured in foliage and, if possible, fine roots to improve understanding of N acquisition and use. Measurement of key parameters associated with plant N metabolism [i.e., plant N pools (NO₃^–, free amino acids and protein), C pools (starch and sucrose)] and with the activity of key enzymes associated with C and N metabolism will provide physiological data on N metabolism and relocation within the plant and will improve characterization of N use by different functional types.
• P concentration in foliage will be measured using a Lachat autoanalyzer and will be evaluated in relation to N concentration as a potential limiting growth factor.

Parameterize nitrogen allocation model. Data collected for Tasks V2 and V3 will be used to parameterize a plant nitrogen allocation model (Xu et al. 2012) that will provide guidance for defining N‑based PFTs and input to CLM on vegetation feedbacks to climate-related changes in soil N availability. The model derives the proportion of carboxylation nitrogen based on temperature, CO₂, and radiation conditions, which is then fed into the Farquhar photosynthesis model. Model output will be compared to calculated values (Task V2.1) based on direct measurement of v_c,max and leaf N concentration. Plant acclimation to climate is simulated by dynamically adjusting nitrogen allocation for light absorption, electron transport, carboxylation, respiration, and storage. The acclimation capability can be different for different species. Nitrogen allocation coefficients are then provided in a look-up table to the Ecosystem Demography model (Fisher et al. 2010), which will be used to track the functional nitrogen availability through simulation time. Optimal nitrogen content per unit leaf area will be estimated in the model by maximizing the nitrogen-use efficiency at the individual leaf level. The estimated optimal area-based leaf nitrogen content will be compared with observed values to assess the capability of plants to adjust their leaf mass per unit area (LMA). Model sensitivity will depend on the plant nitrogen allocation strategies and leaf area-nitrogen dependence. Sensitivity analysis for parameters (e.g., nitrogen storage duration) of different species will be used to define how to group species into PFTs that are responsive to changing N availability.

Initialize PFT representation. The relationships we establish that relate plant community composition to geomorphic units, coupled with larger-scale information of the distribution of geomorphic units within a grid cell, will provide a basis for model initialization of fractional PFT representation. This estimate will not depend on an assumption of linear scaling from point estimates of PFT composition and hence will be a more accurate representation of the complexity of the arctic landscape. Landscape-scale GPP estimates will then emerge from the coupling of fractional PFT distribution with the PFT-specific physiological parameters describing photosynthesis (e.g., v_c,max). By incorporating new functional relationships between PFTs and N dynamics, and N dynamics within geomorphic units, a dynamic vegetation component [e.g., the Ecosystem Demography model (Fisher et al. 2010), being developed as the next-generation vegetation model for CLM] can be introduced to CLM that permits changes in the fractional PFT distribution (and resulting GPP) from the initial condition as permafrost degradation is simulated. Similarly, the model of the plant component of albedo will be initialized by combining the parameter set of albedo of individual PFTs with the fractional representation of PFTs within the grid cell.

Measure carbon flux across scales. The model structure for GPP calculation will be evaluated against observations at both plot and landscape scales. Plot-scale (several square meters) measurements of C flux can be made periodically across the thermokarst gradients to compare short-term model estimates of net C exchange (GPP minus ecosystem respiration) for the given mix of PFTs with the measured flux. These small-scale, instantaneous chamber-based measurements will also enable tests of the model to resolve differences in CO₂ and CH₄ fluxes from different geomorphologic units. The framework for predicting dynamic vegetation in response to permafrost degradation can be tested against measured differences in plant community composition across existing thermokarst features.

• Data-based landscape-scale estimates of GPP cannot be generated by up-scaling plot-scale measurements because of the highly heterogeneous nature of plant distribution and productivity (Street et al. 2007). A preferred approach will be to exploit emergent properties of the landscape that are seen through the relationship between leaf area index and total canopy N content (Williams and Rastetter 1999, van Wijk et al. 2005), which together are strong predictors of photosynthetic capacity and gross primary production (Williams et al. 2001, Ollinger et al. 2008).
• Plot-scale chambers. Portable gas exchange chambers (~ 1 m³) constructed from flexible greenhouse material (Huxman et al. 2004) with a tripod equipped with an open path infrared gas analyzer (LI-COR 7500), air temperature and photosynthetically active radiation sensors, and mixing fans will be used to measure plot-level net CO₂ and H₂O exchange. This approach integrates the fluxes of all the PFTs on each plot. These data will be used to verify the ability of the model to predict C dynamics from information about PFTs.
• Landscape-Scale GPP. Predictive relationships will be established between plant spectral data and the N content and photosynthetic capacity of different PFTs  and combined with observations of plant community composition. Canopy-scale N concentrations will be estimated from canopy N obtained through aircraft imaging spectroscopy, in combination with LAI from remote imagery of NDVI. These estimates will then be used to derive fine-scale estimates of gross primary production through direct comparison with plot and tower-based measurements and to landscape-scale estimates of GPP for model evaluation.

Measure plant contribution to albedo across scales. The composition of the plant community will affect albedo during the snow-free part of the year because different plants (or PFTs) absorb and reflect radiation differently, and emergence of shrubs above the snow cover will affect albedo during the winter (Sturm et al. 2005). The vegetation component of albedo will be estimated from leaf-level spectral data combined with PFT distribution. Unlike most of the other feedbacks between land and atmosphere, albedo can be measured directly at the scale of a grid cell as well as at plot scales and on individual plant leaves, creating strong opportunities for testing our scaling approach. Albedo at the landscape scale will be measured via remote sensing and will be compared with modeled values after the vegetation component is integrated in the model with the albedo from lakes and snow.

Phase 1 Deliverables

• Plant community composition descriptions from study sites for the development of Arctic PFTs.
• Physiological characterization of plant species, including photosynthetic parameters, spectral signatures, and N metabolism, for N allocation model and predictions of albedo and GPP.
• Measurements of plant-available N to develop a predictive model of plant community composition and dynamic N-based PFTs.