Author for correspondence: Colleen M. Iversen Tel: +1 865 241 3961 Email: firstname.lastname@example.org
Experimental evidence from a diverse set of forested ecosystems indicates that CO2 enrichment may lead to deeper rooting distributions. While the causes of greater root production at deeper soil depths under elevated CO2 concentration ([CO2]) require further investigation, altered rooting distributions are expected to affect important ecosystem processes. The depth at which fine roots are produced may influence root chemistry, physiological function, and mycorrhizal infection, leading to altered nitrogen (N) uptake rates and slower turnover. Also, soil processes such as microbial decomposition are slowed at depth in the soil, potentially affecting the rate at which root detritus becomes incorporated into soil organic matter. Deeper rooting distributions under elevated [CO2] provide exciting opportunities to use novel sensors and chemical analyses throughout the soil profile to track the effects of root proliferation on carbon (C) and N cycling. Models do not currently incorporate information on root turnover and C and N cycling at depth in the soil, and modification is necessary to accurately represent processes associated with altered rooting depth distributions. Progress in understanding and modeling the interface between deeper rooting distributions under elevated [CO2] and soil C and N cycling will be critical in projecting the sustainability of forest responses to rising atmospheric [CO2].
Belowground processes are increasingly recognized as an important foundation for ecosystem responses to rising atmospheric CO2 concentration ([CO2]). Fine roots (i.e. roots < 2 mm in diameter) are important in water and nutrient uptake, and are the main interface between trees and the soil ecosystem. Because of their intimate association with the soil profile, fine-root inputs are often more important than leaf litter in driving soil organic matter accumulation (Russell et al., 2004). Rising atmospheric [CO2] is expected to increase carbon (C) and nitrogen (N) allocation to fine roots, especially in N-limited forests (Norby & Jackson, 2000). Increased fine-root allocation could drive changes in soil C storage and N cycling because fine roots turn over quickly in forests (Gill & Jackson, 2000), and contribute a large amount of C and N to the soil system (Iversen et al., 2008).
Experimental evidence from a diverse set of forested ecosystems indicates that fine roots of trees exposed to elevated [CO2] are distributed more deeply in the soil profile relative to trees grown under ambient [CO2] (Table 1). A multitude of important soil properties change with soil depth; for example, oxygen content, soil moisture, bulk density, temperature and soil texture (Schenk, 2005). Thus, as soil depth increases, microbial activity, nutrient availability, and root decomposition rates often decline (Gill & Burke, 2002). While rooting depth distribution under elevated [CO2] was described as a major unknown 15 yr ago (Rogers et al., 1994), the consequences of increased fine-root proliferation and turnover at depth are still poorly understood; this is in part because belowground research is often truncated at relatively shallow soil depths (c. 20 cm). The objective of this review is to examine the potential mechanisms for, and consequences of, deeper rooting distributions under elevated [CO2] as they relate to ecosystem C and N cycling. The main focus is on forest ecosystems exposed to elevated [CO2] in relatively intact soil systems (i.e. free-air CO2 enrichment experiments and open-top chambers).
Table 1. Fine-root (i.e. roots < 2 mm diameter) depth distributions in CO2-enriched woody ecosystems
Experimental manipulation and site location
Target [CO2] (ppm)
Root measurement methodology
Soil depth (cm)
Fine-root depth distribution under elevated [CO2]
Proportion root biomass deeper than c. 15 cm
In addition to the studies reported in the table, there were also several open-top chamber experiments that began in or before 1992 that did not examine rooting depth distribution (e.g. Murray et al., 1996; Norby et al., 1995; Rey & Jarvis, 1997; Tingey et al., 1997; Tissue et al., 1997). In some cases, multiple manuscripts from the same experiment were included when they encompassed different treatment years. Wiemken et al. (2001), Spinnler et al. (2002), Lukac et al. (2003), Day et al. (2006), Liberloo et al. (2006), Johnson et al. (2006), and Phillips et al. (2009) reported the depth distribution of root standing crop, not production. Raw data to determine the proportion of root biomass distributed deeper than c. 15 cm in the soil (actually deeper than 10–25 cm depending on the depth increment measured in the individual study) were obtained from each manuscript from tables, or from figures using digital calipers. In the case of Lukac et al. (2003), the proportional responses of Populus alba and Populus nigra, the species in which the depth distribution was significantly different under elevated [CO2], were averaged. In the case of Liberloo et al. (2006), the proportion of coarse roots deeper in the soil was averaged over all three Populus species. In the case of Wiemken et al. (2001), proportional responses were averaged over calcareous and siliceous soil types. In all cases, I chose the treatment year in which the largest differences in proportional depth distribution between ambient and elevated [CO2] were reported. While potentially useful as a rough comparison among studies, proportional root distribution in the soil profile below 15 cm should be interpreted with caution, given that this will differ depending on the absolute depth of measurements (i.e. 30 cm compared with 100 cm), does not necessarily correspond to the year in which absolute root production at depth in the soil was greatest (i.e. Iversen et al., 2008; Pritchard et al., 2008a), and neglects rooting responses localized to specific soil depths (i.e. Day et al., 2006; Iversen et al., 2008; Johnson et al., 2006).
1Deeper distribution of mycorrhizal production.
2Deeper distribution of coarse roots under elevated [CO2].
3Deeper root distribution with no overall effect of elevated [CO2] on root production.
Evidence for deeper rooting distributions under elevated [CO2]
Deeper rooting distributions under elevated [CO2] have been observed in a variety of experiments and ecosystems, ranging from free-air CO2 enrichment (FACE) experiments in mature forest plantations to tree seedlings and saplings planted in open-top chambers (Table 1). Fine roots developed under elevated [CO2] are not necessarily found deeper in the soil than fine roots developed under ambient [CO2]. Rather, the relative increase in root production under elevated [CO2] is often greatest below c. 15 cm depth, resulting in a larger proportion of root biomass at deeper soil depths under elevated [CO2] (Table 1). For example, in a FACE experiment in a sweetgum (Liquidambar styraciflua L.) plantation, Iversen et al. (2008) found that, over 9 yr, there was a 220% stimulation in cumulative C inputs from fine roots under elevated [CO2] at 45–60 cm soil depth, compared with a 30% stimulation of root C inputs at 0–15 cm depth. At least half of root-derived C and N inputs in this sweetgum plantation were deeper than 30 cm under elevated [CO2]. Pritchard et al. (2008a) found a similar response in a CO2-enriched loblolly pine (Pinus taeda L.) plantation, where elevated [CO2] resulted in a larger stimulation of root production at 15–30 cm depth compared with 0–15 cm depth. Deeper rooting distributions under elevated [CO2] have also been observed in seedlings in pot studies (1.6-m-deep pots; Derner et al., 2005). Of those experiments that examined rooting depth responses to elevated [CO2], 73% found deeper rooting distributions (Table 1).
While rooting depth is functionally determined by species and ecosystem type (Jackson et al., 1996), observations of rooting responses at depth in the soil are limited by the effort applied and the technology used. For example, the minirhizotrons used by Pritchard et al. (2008a) reached to only c. 30 cm, and the authors indicated that deeper tubes were recently installed to determine whether rooting responses deeper than 30 cm exist. Iversen et al. (2008) and Pritchard et al. (2008a) both used minirhizotron technology to determine root dynamics in mature forest plantations, but investigations of root dynamics in other experiments have used methodology ranging from soil coring to in-growth cores (Table 1).
Deeper root distribution under elevated [CO2] appears to be a relatively dynamic response. Root proliferation at depth did not occur in all experiments exposed to elevated [CO2] (Table 1), and when it did occur, it was both dynamic (i.e. occurring in some treatment years and not others; e.g. Day et al., 2006; Iversen et al., 2008; Liberloo et al., 2009) and species-specific (e.g. occurring in two poplar clones, but not a third, in Lukac et al., 2003). Deeper rooting distributions have also been observed under elevated [CO2] without an overall increase in root production (i.e. a redistribution of roots belowground; cf. Johnson et al., 2006). Increased proliferation at depth in the soil has not been limited to fine roots; increased production of mycorrhizas (Pritchard et al., 2008b) and coarse roots (Liberloo et al., 2006) also occurred deeper in the soil under CO2 enrichment.
A historical focus on roots in shallower soils (i.e. the ‘plow layer’) contributes to the fact that rooting depth responses remain unexamined or unreported in many CO2-enrichment studies (Table 1). For example, it was assumed at the start of the sweetgum FACE experiment that the roots had fully occupied the soil volume in the closed-canopy stand (Norby et al., 2004); the subsequent capture of the rooting depth response was largely fortuitous and a consequence in large part of the depth at which the minirhizotron tubes were installed.
Potential causes of deeper rooting distributions under elevated [CO2]
While much work has been done to examine root proliferation in the soil in response to resource patches (reviewed in Hodge, 2004), the causes of increased root proliferation throughout the soil under elevated [CO2] remain relatively unexplored (Pritchard et al., 1999). A conceptual diagram (Fig. 1) may serve as a framework for future hypothesis testing to determine the potential mechanisms for, and feedbacks from, greater root production at depth in a CO2-enriched atmosphere. Deeper rooting distributions under elevated [CO2] are probably related to three factors: increased resource demand as forest production increases in response to CO2 enrichment; increased C available for allocation to root growth; and limited resource availability in shallower soil as a result of increased microbial or plant competition. These three factors will probably interact to control root ‘decisions’ (i.e. Hodge, 2009) that determine root distribution throughout the soil profile.
Increased C allocation to fine roots under elevated [CO2] is mainly observed in nutrient-limited forest ecosystems (Table 1). Further, CO2 enrichment has increased plant demand for nutrient acquisition in a number of forests (Finzi et al., 2007). Roots often proliferate throughout the soil in response to patches of nutrient availability (Prior et al., 2003; Hodge, 2004), and it stands to reason that mining for nutrients is one of the main reasons for greater root proliferation in deeper soil under elevated [CO2] (Fig. 1a). Others have shown that nutrients are available for plant uptake at depth in the soil (Jobbágy & Jackson, 2001; McKinley et al., 2009), and that the proliferation of new roots can stimulate the mineralization of older organic matter (i.e. priming; Dijkstra & Cheng, 2007). However, there is still much uncertainty regarding the cues for root proliferation throughout the soil, as well as the benefits from such proliferation (Hodge, 2004), and little work has been done to examine root proliferation at depth in the soil in response to nutrients. Roots also proliferate in water zones (Hodge, 2004), and greater root production at depth may also occur in response to increased tree water use under elevated [CO2] (Uddling et al., 2008). However, water limitations may be rarer under elevated [CO2] if decreased stomatal conductance at the leaf level (Medlyn et al., 2001) results in less transpiration at the canopy level.
Terrestrial ecosystems are often limited by multiple factors, including light and nutrient availability (Fahey et al., 1998). In nutrient-limited forest ecosystems, greater C fixation in response to rising atmospheric [CO2] may help to alleviate previous constraints on root development and resource acquisition (Pritchard et al., 1999; Stitt & Krapp, 1999). Cost–benefit models have been used to explain root construction and maintenance (Eissenstat et al., 2000), and C gains under elevated [CO2] may shift the cost–benefit balance in favor of root production (Fig. 1b), especially in deeper soil where the benefit of smaller resource gains may have previously been outweighed by C costs. The benefit of root proliferation at depth may be further enhanced by strong competition from microbes, and intra- and interspecific interactions with other plant roots, for limited resources in shallower soil (Fig. 1c), especially as increased litter inputs under elevated [CO2] are expected to increase microbial immobilization of available nutrients (Zak et al., 2000).
Plant root systems are controlled by complex interactions between genetic constraints and environmental conditions (Nibau et al., 2008). Thus, differences in the rooting depth distributions observed under elevated [CO2] across a range of experiments (Table 1) are probably determined by the interplay between genetically determined species characteristics such as plant physiology, biochemistry, and root architecture (Bradley & Pregitzer, 2007; Nibau et al., 2008), and ecosystem properties such as climate and soil texture (Jobbágy & Jackson, 2000), resource heterogeneity (Prior et al., 2003), and water table depth (Imada et al., 2008). For example, this review focuses on forested ecosystems, but, in contrast to forests, crop and grassland ecosystems tend to have shallower rooting distributions under elevated [CO2] (as reviewed in Arnone et al., 2000; Pritchard & Rogers, 2000). Elevated [CO2] has been shown to stimulate the development of lateral roots (Crookshanks et al., 1998; Pritchard et al., 1999). Therefore, root proliferation in shallower soils may be the result of shallower rooting distributions in crop and grassland ecosystems compared with those in forested ecosystems (Jackson et al., 1996), or shallower rooting distributions in annual compared with perennial plants (Holmes & Rice, 1996). Greater access to nutrients or water at shallower soil depths in crop or grassland ecosystems (Prior et al., 2003; Nippert & Knapp, 2007) may also help to explain the contrasting response.
Potential consequences of deeper rooting distributions under elevated [CO2]
While pinpointing the mechanisms of deeper rooting distributions under elevated [CO2] requires more experimentation, the potential consequences of increased root production at depth can be inferred from current knowledge regarding changing ecosystem processes with soil depth.
Deeper rooting distributions under elevated [CO2] may also affect root infection by symbionts (Fig. 1). Mycorrhizal fungi, which receive a significant portion of the C taken up by the host plant in exchange for nutrient uptake, are important players in ecosystem C and nutrient cycling. Further, mycorrhizal abundance has been shown to increase up to 50% in response to elevated [CO2] (Treseder, 2004). Mycorrhizal colonization is closely related to root distribution in the soil across multiple biomes, and, while infection rates tend to decline with soil depth in natural ecosystems (Treseder & Cross, 2006), there is evidence that both ectomycorrhizas and arbuscular mycorrhizas increase root infection rates deeper in the soil profile in response to elevated [CO2] (Rillig & Field, 2003; Pritchard et al., 2008b).
Root inputs and soil organic matter cycling
The vertical distribution of organic matter and nutrients in the soil is strongly related to rooting patterns (Jobbágy & Jackson, 2000, 2001). Thus, the increased proliferation of roots at relatively unexplored depths under elevated [CO2] (Fig. 2) may affect previously stable organic matter pools deeper in the soil. The energy gained in deeper soils from fresh inputs of labile C and N compounds from root exudation (de Graaff et al., 2007), or of detritus from root turnover (Iversen et al., 2008), may be more important than temperature and moisture in stimulating the decomposition of ancient C deeper in the soil profile (Fontaine et al., 2007). For example, rhizosphere priming through exudation by living roots has been shown to stimulate the decomposition of organic matter (Dijkstra & Cheng, 2007), and also stimulate N mineralization (de Graaff et al., 2009). As up to 50% of soil C is stored below 20 cm in forests (Jobbágy & Jackson, 2000), even small changes in C inputs at depth in the soil can have drastic consequences for long-term soil C storage (Fig. 1). However, root exudation is notoriously difficult to measure, especially in situ in the soil, and measurements are often restricted to shallower soil layers (Phillips et al., 2008). Further, more integration is needed to link root C and N inputs with the cycling of organic matter at depth in the soil, where declining oxygen and temperature may be expected to halve microbial activity (Gill & Burke, 2002).
In contrast to the stimulatory effect of fine-root inputs on the decomposition of organic matter at depth in the soil, root-derived inputs have been shown to be disproportionately important for the formation of stable microaggregates in the soil system (Gale et al., 2000). As the process of microaggregate formation depends not only on the organic nucleus of root detritus, but also on soil texture, bulk density, and microbial activity (Six et al., 2002), the rate of formation could be expected to differ throughout the soil profile, though this has not been examined in detail.
Tools and measurements
Novel analyses may be required to determine the consequences of increased root proliferation at deeper soil depths under elevated [CO2] for ecosystem C and N cycling. For example, while minirhizotron measurements are currently the best way to track the dynamics of ephemeral root populations (Johnson et al., 2001), improved methods of extrapolating measurements of root length and diameter obtained from digitized images to root mass and N content (Iversen et al., 2008) and root respiration (Makita et al., 2009) will be key in tracking root-derived C and N cycling at depth in the soil. Other recent tools are also available to track C fluxes throughout the soil that may be attributable to fine roots. For example, the effect of deeper rooting distributions on gradients of soil [CO2] can be determined with CO2 sensors that are coupled with minirhizotron tubes (e.g. Vargas & Allen, 2008). Also, the flux of 13C-CO2 from the soil surface in experiments where root material is labeled with a depleted 13C signal can be measured on diurnal scales with tunable diode lasers (e.g. Bahn et al., 2009). Novel analyses of compound-specific isotopes (Filley et al., 2001) and tissue-specific biopolymers (Filley et al., 2008) may also aid in the identification of root-derived compounds. Along with C fluxes, new strategies are needed to link measurements of soil N availability with root dynamics throughout the soil profile (as reviewed in Frank & Groffman, 2009), as current metrics to examine soil nutrient cycling often exclude the effects of roots.
Incorporating rooting depth in projected forest responses to rising CO2
An important goal of climate change research is the integration of experimental data with ecosystem models (Classen & Langley, 2005). The plant–soil interface is one of the largest areas of uncertainty in current global models, both because of the difficulty in representing complex belowground processes, and also because of the scarcity of data that will allow the development and parameterization of improved model frameworks (Ostle et al., 2009). While the effects of rooting depth distribution on the distribution of C and nutrients in the soil indicate that the interface between fine roots and soil nutrient cycling should be considered throughout the soil profile (Jackson et al., 2000), the most commonly used belowground ecosystem models simulate the mineralization of organic matter at relatively shallow soil depths (i.e. c. 20 cm; Parton et al., 1988).
There are at least a dozen ecosystem or land surface models that are used to project C and nutrient cycling in forest ecosystems (Hanson et al., 2004). The output from these models is an important intermediate step between data obtained from field experiments and the information needed for global models used by the Intergovernmental Panel on Climate Change Report to project future climatic conditions (Denman et al., 2007). However, these models differ in the way in which they represent root distributions and nutrient cycling in the soil profile (Fig. 3). Some models do not have a framework to explicitly consider interactions between fine-root production and soil nutrient cycling (cf. Jackson et al., 2000), and therefore greater root proliferation throughout the soil would not affect nutrient uptake rates by the forest system. While other models contain soil layers that represent different depth increments, and prescribe fractional root allocation among soil layers, these layers (represented by dashed lines in Fig. 3) are typically included to represent water dynamics rather than nutrient dynamics (e.g. Thornton et al., 2007). Decomposition dynamics, and therefore soil C and N mineralization, are modeled as one ‘box’ (heavy outline in Fig. 3), and are typically parameterized with data measured at relatively shallow soil depths (Parton et al., 1988).
A disconnect between observed root dynamics and modeled nutrient availability has confounded projections of forest responses to elevated [CO2]. While models predict that soil N availability will limit forest responses to elevated [CO2] (Thornton et al., 2007), many of the forested FACE experiments found a sustained increase in N uptake from the soil in response to CO2 enrichment (Finzi et al., 2007). There has been much speculation on the source of this ‘extra’ N (Johnson, 2006), and a greater cumulative amount of N available at depth in the soil may be the answer (i.e. a ‘bigger box’ of N when deeper soil depths are considered). However, shallower soil depths have been the main focus of N cycling research in forested FACE experiments to date (i.e. Zak et al., 2003).
A modeling framework that integrates root dynamics and soil N availability at depth may help to pinpoint the cause of increased root proliferation at depth in response to rising [CO2] (i.e. shifting nutrient or water limitations) and explain the dynamic nature of the rooting depth response observed in a number of ecosystems. A framework for examining the interaction among rooting responses and soil C and N cycling at different soil depths currently exists in at least some ecosystem or land surface models (i.e. dashed lines in Fig. 3). However, more data are needed to accurately parameterize root and nutrient dynamics at depth in the soil. Data-model synthesis will be further improved by communication among researchers as to the relevant depth increments for observation of root and nutrient dynamics (Table 1).
Increased root proliferation at depth may be a key response of forested ecosystems to rising atmospheric [CO2]. However, it is uncertain to what extent this is a common phenomenon (Table 1). While I have focused on the responses of woody ecosystems to rising atmospheric [CO2], rooting depth distributions in other ecosystems, such as grasslands and crops, also exert important controls over C and N storage in the soil. Contrasting rooting depth distributions of forested compared with cropped and grassland ecosystems under elevated [CO2] merit further study, and could help to elucidate the proximal controls over altered rooting distributions in a CO2-enriched atmosphere.
Deeper rooting distributions under elevated [CO2], and the interaction of those roots with a soil environment depleted in oxygen and microbial activity, could lead to changes in root form and function, as well as changes in the rate at which root detritus is incorporated into soil organic matter. Altered rooting distributions also provide exciting opportunities for research on C and N cycling in the soil. Advances in the measurement of processes occurring at the root–soil interface that take advantage of novel methodologies such as sensors embedded throughout the soil profile, isotopic partitioning of C fluxes, compound specific chemistry, and measurements of nutrient cycling at depth will provide data needed to inform ecosystem models.
It is important to accurately represent and parameterize processes occurring throughout the soil profile in models. The interactions among elevated [CO2] and global change factors such as rising temperatures and altered precipitation regimes will almost certainly affect the responses described here, albeit in uncertain ways. Furthermore, the reconfiguration of current model frameworks to accept data on rooting distributions and nutrient cycling at depth in the soil will facilitate the testing of new hypotheses such as those conceptualized in Fig. 1. Continued progress in understanding the interface between root growth and turnover and soil C and N cycling, especially at depth in the soil, will provide critical information needed for understanding current ecosystem function, as well as predicting future ecosystem responses to environmental change.
Thank you to M. A. de Graaff, P. Hanson, R. Norby, J. Warren and three anonymous reviewers for comments that improved an earlier draft of the manuscript. Research was supported by the United States Department of Energy, Office of Science, Biological and Environmental Research. Oak Ridge National Laboratory is managed by UT-Battelle, LLC for the United States Department of Energy under contract DE-AC05-00OR22725.