Introduction to a Virtual Special Issue: modeling the hidden half – the root of our problem


Fine roots, together with their microbial associates, sustain the high nutrient and water demands of photosynthesis, a crucial process for life on Earth. They are also a critical building block for soil organic matter (SOM) formation and stabilization (Jastrow & Miller, 1998), and they participate in the weathering of silicates (Berner, 1998). These processes play vital roles in the biogeochemical terrestrial carbon, nutrient, and water cycles.

Although fine roots interact with microorganisms and the soil environment, influencing and supporting ecosystem processes and properties, they are difficult to observe, and their activities are not easily measured. The ‘hidden half’ ecosystem (Waisel et al., 1991) represented by roots is a long-term concern in ecosystem science and biogeochemical cycling. In the mid-1990s, belowground processes were identified as a key area where fundamental knowledge was lacking: (1) for predicting changes in terrestrial ecosystem functions and properties in response to climate forcing factors; and (2) as an aid in interpreting ecosystem feedbacks (Curtis et al., 1994; Canadell et al., 1996). The New Phytologist special issue ‘Root dynamics and global change: an ecosystem perspective’ (Norby et al., 2000) was a decisive step toward identifying both advances in root ecology and significant research needs. Studying the root as a dynamic system that might be impacted by environmental changes and adequately representing roots in global models were found to be major needs for understanding the carbon cycle.

‘The “hidden half” ecosystem (Waisel et al., 1991) represented by roots is a long-term concern in ecosystem science and biogeochemical cycling.’

More than a decade later, significant gaps remain in our fundamental understanding of root systems and their dynamic interactions with broader ecosystem-level processes. At the ‘Scaling root processes: global impacts’ Workshop, held in Washington, DC (USA), on March 7–9, 2012, more than 50 belowground researchers gathered to identify new research approaches and technologies for improving fundamental understanding and model parameterizations of root processes in predictive climate models. At this workshop, the experimental and modeling communities began a dialog on the current representation of root processes in models and the needed improvements. The concept that large-scale models work well with only rudimentary root system functionality (or none) was also introduced. This unsettling fact has limited motivation to seek improvement in models by adding root functions.

It was eye-opening to see that although root distributions in soils and root chemistry were simulated to an extent, neither structural–physiological root functions nor root responses to the soil environment were incorporated into large-scale models, such as the terrestrial components of Earth system models (ESMs) like the Community Land Model (CLM). In a fundamental way, roots are treated as a passive continuation of the aboveground leafy system, and most root physiological parameters are tied to aboveground dynamics, with water and nutrient uptake determined by plant demand and an invariable root system. In the soil, fine roots entered litter pools, just like leaf litter.

This revelation raises very puzzling and challenging issues. How could models work well without root functions? Even though models did not explicitly incorporate root functions, other plant or soil parameters could be tuned to simulate ecosystem responses related to root structure and physiological activities (Leuzinger & Thomas, 2011). However, how confident could we be in future predictions if belowground mechanisms and drivers were not explicitly parameterized in models, particularly in the face of environmental changes that can affect aboveground and belowground environments differently? If root functionality were incorporated into models, this would at least add more realism and might enhance simulations and predictive capabilities in a changing world.

Given all of the advances in root research and current awareness of the importance of belowground drivers of ecosystem processes (McCormack & Fernandez, 2011), an appropriate question is: why have root processes not been incorporated into models? The articles in this Virtual Special Issue (VSI), many of which were prompted by the workshop discussions, discuss different aspects of root structure and function and their interaction with the soil and microbial environment, with a focus on processes and mechanisms that should be incorporated into ESMs. These articles challenge some common assumptions used to represent root physiological functions, identify root–microbial–soil environment processes that could be incorporated into models, and demonstrate how new methods and modeling approaches can improve the representation of physiological function of roots in models.

Functional equilibrium in carbon–nutrient Interactions

Valentine & Mäkelä (2012), Bahn et al. (2013), McMurtrie & Dewar (2013), and Smithwick et al. (2013) challenged the assumption of functional equilibrium allocation between shoots and roots to determine root growth used by most models. Valentine & Mäkelä (2012) showed emergent properties of the optimization model, revealing complex behaviors of fine roots to nitrogen availability and stand age. Bahn et al. (2013) tested assumptions by shading grass species and showed a preferential carbon flow to belowground plant functions (respiration and storage) and to fungal communities and rhizosphere microbes, at the expense of the carbon status of the aboveground organs. This is opposite to the result expected from the functional equilibrium hypothesis. Also, for plants growing under elevated CO2, Sonderegger et al. (2013) showed positive feedback effects of roots on shoot growth by means of increased ability to break dormancy and initiate growth, allowing for earlier access to resources during springtime. Similarly, both Smithwick et al. (2013) and McMurtrie & Dewar (2013) showed that the assumption of functional equilibrium allocation is too simplistic in its formulation and is biased toward an aboveground-driven view of root responses. These authors proposed new frameworks, expanding the equilibrium ideas by incorporating root factors (root life span, root production dynamics, mycorrhizal influences on root resilience) influencing aboveground function and allowing for more equal footing between aboveground and belowground responses to stressors. This approach needs to be extended to plants limited by resources other than nitrogen, as well as to nonwoody plants, competition between plants, and soil environmental stresses (soil nitrogen saturation, etc.). Further, Freschet et al. (2010) provided evidence of high nutrient resorption for both leaves and roots of arctic plants, suggesting that nutrient resorption is highly important for the plants’ nutrient budget, independently of environmental factors, and is influenced by multiple physiological parameters.

For instance root respiration is currently modeled by using passive climate sensitivity assumptions that have been challenged by empirical data. Hopkins et al. (2013) propose changes in the traditional modeling of root respiration, providing ecosystem-scale evidence that substrate availability (and hence photosynthesis) can exert greater control than soil moisture and temperature on root respiration. They also propose the incorporation of root respiration as a primary function of gross primary production, responding to environmental variables by shifting carbon allocation belowground, with consequences for exudation and soil microbial activity. In addition, Bloemen et al. (2013) demonstrate the degree of complexity of root respiration in cottonwood trees by showing that as much as 47% of root-respired CO2 was transported to aboveground wood tissues and afterwards lost to the atmosphere, while c. 17% of the transported root-respired CO2 was assimilated into woody and leaf tissues. Both phenomena need to be incorporated into models.

Root structural–physiological dynamics

Data emerging from recent advances in measuring and interpreting root turnover challenge current simulations of annual root turnover in models. Xia et al. (2010) suggested that fine roots of woody species are complex branching systems with both rapid-cycling and slow-cycling components. The functions of fast-cycling or ephemeral roots and more permanent roots are different. Also, Keel et al. (2012) showed that functional differences in the fine-root population were indicated by carbon allocation patterns and residence times. They found that allocation of recent carbon and higher turnover rates in root tips where associated with their role in nutrient and water acquisition. Roots with longer mean residence times but high allocation to sugars and starch had a role in structural support and storage. Riley et al. (2009) proposed the Radix model to separate fast and slow pools in root carbon turnover, on the basis of changes in carbon-14 (14C) enrichment and dilution in live and dead root populations after an entire forest was enriched with 14C. Lynch et al. (2013) employed a whole-ecosystem carbon-13 (13C) tracer to consider the implications of differential root turnover for SOM accumulations. The authors compared a single-pool model to a two-pool model representing a heterogeneous root population, for simulating root carbon inputs to soil vs measured rates of soil organic carbon accrual under elevated CO2. Consideration of two root populations (as in the Radix model) improved modeling of soil carbon accrual and doubled the fine-root contribution to forest net primary production. Overall, these studies show that accounting for heterogeneity in carbon residence times in fine roots will improve and probably constrain the estimates of fine-root production. The need to find proxies for determining root dynamics (production, life span, etc.) has produced several easily measurable parameters that correlate with root function. Ontl et al. (2013) tested the effects of topography and soil properties on root production and found good correlations for one species. Thus, they suggest the use of existing spatially explicit soil databases for estimating root production for such species. In addition, McCormack et al. (2012) considered the possibility of using a combination of whole-plant and root traits to determine fine-root life span whenever empirical data are lacking. They found good correlations between root life span and tree diameter at breast height, root diameter, and root carbon-to-nitrogen (C : N) ratios. This approach could be implemented to determine root longevity and root inputs to soil in large-scale models.

Root–soil interactions

Some of the greatest modeling challenges are those root functions that are related to interactions with the soil environment. Prieto et al. (2012) and Neumann & Cardon (2012) provided reviews on the state of knowledge and modeling approaches for representing the hydraulic redistribution (HR) of water through the root system. The passive HR process, which is driven by water potential gradients between soil layers and the root system, can move water upwards or downwards through the root and to the soil. Prieto et al. (2012) highlighted the global implications of HR, acting at the plant level but influencing the structure and function of plant communities. Neumann & Cardon (2012) reviewed modeling approaches to represent HR. They pointed out that when HR is used in large-scale models to improve the match between modeled and measured transpiration rates, the model results in a lack of HR functionality. The cause, in part, could be a difficulty in parameterizing and validating HR in large-scale models due to limited information on plant–root and soil characteristics. The neutron tomography (NT) technique (though not a field method) can quantify and visualize water content in soil with high spatial and temporal resolutions; NT could provide valuable information for parameterizing HR in models. Moradi et al. (2011) used NT to study the water content in the rhizosphere and bulk soil for plants growing in columns of sandy soil, and provided evidence that the rhizosphere soil holds more water than bulk soil, thus improving soil hydraulic properties and facilitating water and nutrient uptake at low soil water contents.

‘Some of the greatest modeling challenges are those root functions that are related to interactions with the soil environment.’

Root functions that could be incorporated into models include associations with mycorrhizal fungi and soil functions. For example, Kuzyakov & Xu (2013) introduce the concept of ‘cooperation’ between plants and microbes. In this review, the authors establish that competition for nitrogen uptake between plants and microorganisms is facilitated at the rhizosphere, where root density and mycorrhizal formation are crucial in shifting nitrogen acquisition to roots and preventing nitrogen losses, with ecosystem-scale consequences. Similarly, Phillips et al. (2013) consider how differences in forest composition (trees and associated microbes) influence biogeochemical processes and how these dynamics can be used to improve representation of plant–soil feedbacks and nutrient constraints on productivity in ecosystem and larger-scale models. These authors propose a Mycorrhizal Associated Nutrient Economy (MANE) framework for predicting the magnitude and direction of forest responses to global change forcing factors, such as nitrogen deposition and elevated CO2. A detailed explanation of how this framework could be incorporated into ESMs shows possible improvements in the representation of C : N ratios across temperate forests, as well as impacts on SOM decomposition and carbon and nitrogen allocation in temperate forests. Other root functions not incorporated into models provide mechanistic understanding of the way roots interact with soil functions through the rhizosphere priming effect (RPE) (Cheng et al., 2013). Incorporation of the RPE was necessary to explain increases and decreases in decomposition of SOM. By incorporating RPE into the PhotoCent model, a 40% increase in decomposition rate, driven by increases in root carbon inputs to the slow SOM pool, reduced SOM content in a forest exposed to elevated CO2, and this result was consistent with observations (Cheng et al., 2013). Without this important root process, the same model would have predicted an increase in SOM — contradicting observations. These studies show that the incorporation of root processes into models can improve understanding of observations and reduce uncertainty in predictions.

New methodologies and new models are needed for observing and understanding functions of the root system. For example, Iversen (2010) provided evidence that without better representation and parameterization of root distribution in the soil profile and its forcing by various factors, the ecosystem carbon cycle will not be accurately captured in models. She proposes a number of methodology and model changes to improve representation of rooting depth distributions. Allen & Kitajima (2013) provide in situ, high-frequency observations of mycorrhizal fungi dynamics and roots, including determination of mortality and life span of mycorrhizas and an assessment of their influence on root and soil surface responses such as soil respiration. Also, Paterson et al. (2009) show how isotope techniques are providing information for understanding mineralization controls on SOM and for linking soil microbial community structure to function. Furthermore, improved models, such as the one produced by Zygalakis et al. (2011) – the ‘dual porosity’ model, which incorporates root geometry and soil microstructure to represent nutrient uptake by root hairs – could also be used to predict changes in nutrient uptake due to environment and forcing factors.

A roadmap for bringing roots into models

The articles included in this VSI present a foundation and justification for which major root functions and associated mechanisms should be incorporated into models, as well as presenting a general roadmap as to how these parameters could be incorporated. A common theme throughout the workshop was the ongoing need to couple experiments and models together to identify critical research areas in root ecology. Areas that need more research include: (1) finding scalable mechanisms that link plant productivity to belowground fluxes and the dynamics of roots, arbuscular mycorrhizal fungi, ectomycorrhizal fungi, and other microorganisms; (2) finding quantitative and mechanistic approaches for studying root populations (and mycorrhizal hyphae) that differ in functionality and turnover rates; (3) studying impacts of root–microbial interactions on the turnover and size of SOM pools; and (4) coupling between hydrology and biogeochemical patterns of nutrient and carbon flow through roots and soil.


The authors thank the Department of Energy for sponsoring the Workshop held in Washington, DC (USA), March 7–9, 2012 ‘Scaling root processes: global impacts’ and NSF for supporting some of the students attending the workshop. The workshop was organized by R. Matamala, B. A. Drewniak, D. M. Eissenstat, M. A. Gonzalez-Meler and C. M. Iversen. Thanks also go to Richard Norby for editorial assistance.