New approach for capturing soluble root exudates in forest soils


*Correspondence author. Department of Biology, Indiana University, Bloomington, IN 47405, USA. E-mail:


  • 1Soluble root exudates are notoriously difficult to collect in non-hydroponic systems because they are released in a narrow zone around roots and are rapidly assimilated by rhizosphere microbes. This has substantially limited our understanding of their rates of release and chemical composition in situ, and by extension, their ecological significance.
  • 2Here we describe the advantages and limitations of several commonly employed methods for measuring exudation with respect to their potential adaptability for field use in forest ecosystems. Then, we introduce a novel in situ method for measuring exudation in forest soils, and present preliminary results of the spatial and temporal dynamics of loblolly pine (Pinus taeda L.) exudation at the Duke Forest FACTS-1 site, North Carolina, USA from April 2007 to July 2008.
  • 3Exudation rates varied by an order of magnitude, with the highest rates occurring in late-June 2007 and mid-July 2008, and the lowest rates occurring during late-August 2007. On an annual basis, we estimate pine roots in the upper 15 cm of soil release c. 9 g C m−2 year−1 via this flux, which represents 1–2% of net primary productivity at the site.
  • 4The magnitude of exudation rates did not differ across an N availability gradient but did track general patterns of below-ground C allocation at the site. Exudation was well-predicted by root morphological characteristics such as surface area and the number of root and mycorrhizal tips, further supporting a possible link between root C allocation and exudation.
  • 5Because all methods for estimating exudates introduce experimental artefacts, we suggest that only a limited amount of ecologically relevant information is probably gleaned from a single method. Thus, a complementary suite of experimental approaches will best enable researchers to understand consequences of changing patterns of exudation in the wake of global environmental change.


Because they occur in a narrow zone of soil around roots and are rapidly assimilated by soil microbes, root exudates are one of the most poorly quantified components of the below-ground C cycle (Wardle 2002; Paterson 2003). Root exudates are soluble, low molecular weight organic compounds that can be passively released to soil due to the concentration gradient between root cells and soil solution or which are actively secreted in response to metal toxicity, nutrient stress and the presence/absence of plant and microbial taxa (Marschner 1995; Jones et al. 2004; Bais et al. 2006). Most exudates consist of sugars, amino acids, and organic acids (Neumann & Romheld 2001), and this flux is believed to represent between 1% and 10% of net assimilated C (Jones et al. 2004). Despite the relatively small magnitude of this flux, root exudates are believed to play an important role in mediating soil nutrient availability in ecosystems due to their chelating properties and their role in stimulating microbial activity (Lynch 1990; Marschner 1995). Furthermore, exudates are primarily derived from recently-assimilated photosynthate (Neumann & Romheld 2001) and thus, may represent a semi-continuous input of labile C to soil in contrast to transient inputs of C resulting from leaf litter inputs (Kuzyakov & Cheng 2001).

Significant methodological challenges limit our ability to accurately measure exudation or to determine how environmental variables affect exudation composition or rates. In general, all exudation methods attempt to overcome a set of common challenges: (i) capturing exuded C before microbial assimilation, (ii) selecting a medium that does not affect root physiology and exudate recovery, and (iii) distinguishing exuded compounds from other soluble C compounds in the medium. Such challenges become even more formidable when adapting a method for field use. In most cases, roots need to be temporarily removed from the soil to be studied, which may stress or injure the root and disrupt mycorrhizal networks (Neumann & Romheld 2001). Moreover, root and rhizosphere processes are highly variable in space and time (Hinsinger et al. 2005), which pose challenges to developing experimental protocols that capture this variability, and to scaling measured values to the ecosystem-scale.

Not surprisingly, there have been few measurements of exudation from trees in situ (but see Smith 1976). This is particularly true for forest ecosystems where deep roots are difficult to access and whole-system isotopic tracers are difficult to employ. Exudation measurements of tree seedlings grown in controlled experimental systems suggest that differences in biotic factors (e.g. plant species, phenology, mycorrhizal status), abiotic factors (e.g. soil fertility, moisture, temperature), and the experimental system employed may all influence the rates and composition of exudates (Grayston et al. 1996). This has led many researchers to conclude that exudation in forest ecosystems should be placed into a ‘black box’ of soil processes and estimated through modelling approaches (Luo et al. 2001) or as residual terms in mass-balance calculations of below-ground C flux (Fahey et al. 2005; van Hees et al. 2005). However, such approaches contribute little to our understanding of the mechanisms that control this process, and by extension, the potential role of exudates in mediating microbial activity, nutrient transformations, and feedbacks to primary productivity and ecosystem C storage (Cheng 1999; Phillips 2007).

Below-ground processes such as exudation mediate the flux of energy and materials in terrestrial ecosystems, but our understanding of how such processes influence feedbacks to ecosystem C and N cycling remains limited owing to significant methodological obstacles. The goals of this paper are to: (i) review existing methods for collection of exudates, (ii) describe a newly-developed method for collecting exudates from tree roots in situ, and (iii) present data collected over a 15 month period from trees at the Duke Forest Free Air CO2 Enrichment Experiment.

existing exudation methods

Several different approaches have been used to measure exudation, with the specific techniques employed depending upon the investigators’ primary research interests. In studies where the goal is to examine the total amount of rhizodeposition (sloughed cells, mucilage, etc.) entering soil and the movement of this C into soil pools, isotopic labelling approaches have generally been employed because non-sterile soils can be used (reviewed in Paterson et al. 1997). In studies of the quantity and chemical composition of exudates, culture-based systems have generally been used because the exudates can be more readily trapped and separated from the medium (Jones et al. 2004).

Two primary types of culture-based systems are static and percolating (i.e. non-static) trap solutions. In general, both systems are similar in that root systems are submerged in a medium from which exudates are trapped and collected over a set period of time. The primary difference between the two systems is the replenishment of the trap solution in the percolating system which maintains the diffusion gradient between root cells and trap solution, and minimizes the re-uptake of exuded sugars and amino acids by roots. Thus, an important first step in employing a static trap solution is deciding when to sample the trap solution. The challenge is to sample the solution after a sufficient amount of C has been exuded (i.e. to reduce signal to noise artefacts), but before exudation rates are affected by C accumulation in the trap solution and re-uptake of exudates by roots (Jones et al. 2004; Personeni et al. 2007). An advantage of percolating solutions is that there is no bias associated with exudate accumulation affecting efflux or influx of compounds, and thus percolating systems may better reflect a rhizosphere environment where microbes are present. However, percolating solutions have a disadvantage in that they generally require a much greater volume of solution, and several post-collection steps may be necessary to for subsequent chemical analyses. In addition, percolating solutions may not uniformly collect exudates released by roots if the solution follows preferential flowpaths in the medium. Because percolating solutions require the continuous pumping of solution through the medium, they may also be less amenable for adaptation to field studies.

In both static and percolating systems, the type of growth medium selected is known to have important consequences for root growth, architecture and exudation. Pure solution cultures (generally a dilute salt or nutrient solution) offer simplicity in maintenance and sample collection. However, the lack of mechanical impedance for roots in non-solid media may affect root morphology and exudation rates (Neumann & Romheld 2001). Small glass beads and acid-washed sand have commonly been employed but such media have limitations as well, as sorption of exuded compounds to these media may result in incomplete recovery of exudates or mischaracterization of the exudate composition (Neumann & Romheld 2001). In our own experiments, we have found that acid-washed sand (3 m HCl) can be both a source and sink for C (R.P. Phillips, unpublished data). Thus, careful consideration should be given to selecting an appropriate medium for the trapping solution as small amounts of sorbed C or contamination of C from the medium can introduce substantial artefacts to estimated exudation rates and compositional analyses.

An important consideration in adapting either trapping system for field use is the question of sterility of the culture medium. Sterile/axenic culture systems have an advantage in that exudates released from roots will not be metabolized by microbes in the medium, thereby allowing exudation rates to be measured without needing to account for microbial assimilation of released of compounds (Uselman et al. 2000). However, the presence of microbes may stimulate exudation by maintaining a concentration gradient around roots or by inducing exudation via the release of signalling compounds (reviewed in Grayston et al. 1996; Paterson et al. 1997; Neumann & Romheld 2001). Using a sterile medium may thus reduce exudation rates and affect the composition of exudates. The issue of sterility is even more vexing for studies in forest soils as most roots are likely to be extensively colonized by mycorrhizal fungi and rhizoplane bacteria. One possible method for minimizing artefacts associated with non-sterile conditions is to treat roots with antibiotics. However, the efficacy of any antibiotic treatment will likely depend on the type and concentration of antibiotics used as well as the plant species of interest (Neumann & Romheld 2001).

In addition to trapping solutions, localized sampling techniques have been used to collect exudates. Agar, specialized resins or filter paper can be placed along desired sections of the root axis to collect compounds released from distinct areas of the root (Neumann & Romheld 2001). These techniques have the benefit of providing information on fine-scale spatial patterns of exudation, and can be coupled to experiments with bacteria cloned with reporter gene systems (Killham & Yeomans 2001; Cardon & Gage 2006). Non-sterile conditions are less likely to bias results, as exudates trapped by resins will be unavailable to soil microbes. However, use of resins, filters, etc. may not be ideal when looking for unknown compounds, as the a priori choice of resin will influence which compounds are trapped. An additional caveat is that exudation rates cannot be readily quantified. Thus, application of this method for field studies might be appropriate in cases where exudate composition, rather than rates, is of interest. Localized sampling of exudates can also be performed using microsuction cups as tension lysimeters (Gottlein & Blasek 1996; Sandnes, Eldhuset & Wollebaek 2005). This method has advantages in that the exudates are removed from soil before they can be metabolized by microbes, and the cups can be repeatedly sampled to examine temporal trends. However, a major limitation of this method is that only small volumes of solution can be collected under most field conditions (e.g. irrigation is often necessary), and flux rates cannot be quantified. Moreover, this method must be employed through rhizoboxes or root windows (Dessureault-Rompre et al. 2006; Shen & Hoffland 2007), and thus only a small volume of soil is generally sampled.

In this manuscript, we describe our use of a culture-based system for measuring exudation from tree roots which is relatively inexpensive, easy to deploy, and requires minimal maintenance. In addition, this method can account for the spatial heterogeneity and temporal dynamics of forest soils and root systems, and the varying degrees of colonization of tree roots by a diverse assemblage of mycorrhizal fungi – all of which would presumably affect exudation rates and the chemical composition of the C released (Grayston et al. 1996; Scott-Denton et al. 2006; Van Scholl et al. 2006). Moreover, exudation rates can be scaled to the ecosystem-level using this method so that the consequences of this flux for microbial activity and nutrient cycling can be better quantified.

Materials and methods

site description

The Duke Forest-Atmosphere Carbon Transfer and Storage (FACTS-1) was established in a loblolly pine (Pinus taeda L.) plantation in Orange County, North Carolina (35°58′N, 79°05′). The site is dominated by the pine (> 90% of the basal area), although hardwoods have developed in the understorey since the initial planting in 1983. In 1996, replicated 30 m diameter plots containing c.100 trees per plot were established at the site. Three experimental plots at the site are fumigated with exogenous CO2 to maintain an atmospheric concentration c. 200 ppmv above ambient levels (i.e. c. 585 ppmv) while three plots are fumigated with air only (i.e. c. 385 ppmv). The mean annual temperature of the site is 15·5 °C and mean annual precipitation at the site is 11 140 mm year−1. Soils at the site (Enon Series) are highly-weathered clay loams (mixed thermic Ultic Hapludalfs) developed from igneous parent materials, and are moderately acidic (pH = 5·6). In general, the vast majority of the roots (> 90%) reside in the upper 30 cm of soil. Detailed descriptions of the experimental set-up and the site characteristics can be found in Hendrey (1999) and Lichter et al. (2005), respectively.

Average daily temperature and photosynthetically active radiation (PAR) were collected during the days of exudate collection from April 2007 to July 2008 (Table 1). These meteorological variables are strongly linked to C assimilation by loblolly pines at the site (Hendrey et al. 1999). Air temperature was collected at 1 min intervals at a mid-canopy location in the middle of each plot using shielded thermocouple with an accuracy of c. 1 °C. PAR data were collected at 1 min intervals using a LiCor Quantum sensor (Licor Co., Lincoln, NE) mounted above the forest canopy. Average daily temperature and PAR were calculated from hourly means for each day of exudate collection, and then averaged for each 3–4-day collection period (Table 1).

Table 1.  Meteorological data from Duke FACTs-1 site, NC on the days of exudate collection. Values are weekly means of daytime averages in air temperature (°C) in the middle of the forest canopy and photosynthetically active radiation (PAR; µmol m−2 s−1) at the top of the canopy

exudate collection

Soluble root exudates were collected from loblolly pine trees at 4–8-week intervals from April 2007 to July 2008 using a modification of a static culture-based system which was adapted for field use (Fig. 1). In this system, intact fine root systems were washed free of adhering soil and ‘incubated’ in glass cuvettes filled with sterile glass beads and C-free nutrient solution. Exudates released by roots were collected over short time-intervals by flushing each cuvette with a vacuum pump. Similar designs using bead-filled cuvettes have been used to collect root and mycorrhizal exudates from seedlings of Pinus sylvestris (Ahonen-Jonnarth et al. 2000; Van Scholl et al. 2006), Picea abies [L]. Karst. and Betula pendula Roth (Sandnes et al. 2005) and loblolly pine (Phillips et al. 2008). Similar static bathing solutions have been used to measure nutrient uptake of pine trees in the field (Lucash et al. 2005). However, this study is the first study to our knowledge to use this type of system to measure root exudation for field-grown trees.

Figure 1.

Sketch of an experimental system employed to trap root exudates from intact loblolly pine roots growing at the Duke Forest FACTS1 site, North Carolina, USA. Note that all root systems are still attached to the tree during exudate collection and only appear unattached for illustrative purposes.

excavation and cuvette assembly

Terminal fine root systems (< 2 mm diameter with laterals) were carefully excavated from soil (sensu Lucash et al. 2005) at the interface between the O and A horizons within 0·5 m from the bole of each tree. In order to ensure that roots were loblolly pine, all root systems were traced back to a parent tree or coarse root with characteristics unique to loblolly pine (e.g. reddish flaky bark). Once unearthed, root segments (15–20 cm length) were rinsed with a squirt bottle containing nutrient solution (to minimize osmotic stress to the root) to remove adhering soil. Fine forceps were also used to dislodge adhering soil. Given the intimacy between the roots, mycorrhizal hyphae and organic matter, we took extreme caution to remove soil during this process, resulting in average processing time of c. 1 h per root system per person. The intact root systems were placed into a soil-sand mixture (1:1) and re-buried until they could be placed into the root cuvettes (generally 24–48 h later). The re-burial was intended to allow additional time for roots to recover from any potential injury or stress sustained during the excavation and rinsing process. Sieved soil mixed with sand was used to facilitate removal of the soil (by squirt bottle) before placement into the root cuvettes.

Soil-free root systems were placed into a 30-mL glass syringe (Popper & Sons, New Hyde Park, NY) from which the plunger was removed, and each syringe was back-filled with sterile acid-washed glass beads (c. 750 µm diameter; La De Da Designs, Baton Rouge, LA). At the narrow end of each syringe, a 30 µm mesh cloth was folded into a cone to support the beads and to prevent them from clogging the syringe outlet during removal of the solution via a vacuum pump (Fig. 1). At the top of each syringe, a rubber septum with a small slit cut to accommodate the protruding root was used to seal off each cuvette. To protect the exposed portion of the root from drying out, a moist Kimwipe (Kimberly-Clark Corp., Roswell, GA) was placed around the upper root segment and secured with Parafilm (3 m Company, Minneapolis, MN). A small volume of dilute nutrient solution (see below) was added to the cuvette to maintain humid conditions during the incubation. The cuvettes were then covered in aluminum foil, returned to the excavated area, and covered with several layers of needle litter to allow the root system to equilibrate with the cuvette environment. Three control cuvettes filled with glass beads only were similarly covered and buried in soil (one in each plot).

exudate collection

After a 2–3 day equilibration period, a dilute nutrient solution was added to each glass cuvette to facilitate the removal of accumulated exudates. In order to ensure the complete removal of C from the cuvette before the experimental incubation period, each cuvette was filled to saturation with C-free nutrient solution (0·5 mm NH4NO3, 0·1 mm KH2PO4, 0·2 mm K2SO4, 0·2 mm MgSO4, 0·3 mm CaCl2) and flushed with a vacuum pump. This process was repeated five consecutive times. After the fifth flush, a small volume of nutrient solution was added, and the root-filled cuvette was re-buried as before. At the end of each incubation period, cuvettes were flushed three times with a C-free nutrient solution to remove accumulated exudates using the procedure described above. We found that three flushes were sufficient to remove over 90% of the soluble C and N in each cuvette in a separate recovery experiment using spiked solutions of low and high C and N concentration (Fig. 2).

Figure 2.

Recovery of soluble C and N from bead-filled cuvettes containing low and high concentration solutions. Panels on the left indicate the total amount of C (a) and N (c) flushed from cuvettes containing a low concentration solution, while panels on the right indicate the total amount of C (b) and N (d) flushed from a high concentration solution.

Following each incubation period, all cuvettes were reopened and glass beads rinsed from roots. We did not analyse the beads following their removal after preliminary tests suggested little residual C accumulation on the beads (data not shown). Washed roots were removed from the tree, photographed, imaged, and analysed for root morphological characteristics using WinRhizo (Regents Instruments Inc., Québec, Canada). All solutions were filtered immediately through a sterile 0·22 µm syringe filter (Millex PVDF, Millipore Co., Billerica, MA), and refrigerated at 4 °C until analyses (< 48 h). All samples were analysed for non-particulate organic C and total N on a TOC analyser (Shimadzu Scientific Instruments, Columbia, MD). Only the C analyses are presented in this article.

experimental tests

Experiment 1: Exudate accumulation and incubation duration

An important consideration for static bathing solution culture-systems is the degree to which exudation rates are affected by the changing C concentration in the trap solution and the potential re-uptake of exudates by roots (Jones et al. 2004; Personeni et al. 2007). We initiated a pilot experiment to examine this potential source of bias using loblolly pine roots from two plots (one exposed to ambient CO2, the other elevated CO2) at the Duke Forest FACTS-1 site in July, 2007. We collected exudates at 0, 4, 8, 16, 21, 24 h from a total of eight root systems (two per tree) over four consecutive days (July 17–20). Because of the time needed to flush and extract the solutions, not all eight root systems could be incubated for the same time intervals on a given sampling day. Thus, we staggered our sampling so that subsets were incubated on different days. This also enabled us to minimize the potential influence of diurnal variation on exudation (Neumann & Romheld 2001). Root-free controls were flushed at the same time intervals to serve as controls.

Experiment 2: Day to day variation in exudation rates

Given the potential bias of sampling root systems on different days due to variation in environmental conditions (e.g. temperature, PAR, wind speed etc.), we initiated a second experiment at FACTS-1 to examine the variation in exudation rates over consecutive days. We tested this potential source of variation by incubating 11 root systems in two plots (one at ambient CO2, the other at elevated CO2) for 24 h intervals during consecutive days in July and September 2007. The 24 h incubation time was chosen after preliminary tests using loblolly pine roots revealed that exudation rates were still increasing after 24 h (see Experiment 1 results below) and were of sufficient magnitude to be analytically detectable. In both samplings, cuvettes were flushed five times between dates to ensure that no exudates remained from the incubation of the previous day. Exudates were collected with three flushes and analysed for non-particulate organic C on a TOC analyser (as described above).

Experiment 3: Seasonal variation in exudation rates

We sought to quantify pine root exudation rates at the FACTS-1 site across a range of environmental conditions. Thus, we initiated an experiment to collect exudates every 4–8 weeks from April 2007 to July 2008 in three experimental plots (all at ambient CO2) which vary in annual net N mineralization rates by approximately a factor of three (Finzi & Schlesinger 2003). All exudates were collected from two or more root systems in each of the three plots – with the exception of the April, June and July 2007 when exudates were collected from root systems in a single plot. All of the incubations for this experiment lasted for c. 24 h, beginning and ending in the early hours of the photoperiod to account for potential biases associated with diurnal variation in exudation (Neumann & Romheld 2001). All exudates were collected and processed as described above.

calculations and statistics

In all three experiments, C accumulation in each cuvette was calculated as the sum of the three flushes minus the total C flushed from the root-free control cuvettes (units of µg C). For experiment 2, Pearson correlation coefficients were used to examine day to day variation in C efflux from 11 root systems. In addition, a paired t-test was used to evaluate whether the paired means were significantly different from one another (α = 0·05). For experiment 3, we used analysis of variance (anova) to examine differences in exudation rates across the N availability gradient of the three plots (α = 0·05). We also used anova to examine monthly and seasonal variation in exudation (n = 6 on each sampling date), and linear regression to examine the relationship between C efflux and root morphological variables across seasons. Seasons were defined relative to the dates samples were collected: spring (April 2007 and May 2008), summer (June, July and August 2007, July 2008), fall (September and October 2007), and winter (December 2007, February and March 2008). Non-normally distributed data were transformed before statistical analyses with jmp statistical software (v. 6, SAS Institute Inc., Cary, NC).


experiments 1 and 2

We evaluated whether C accumulation in the cuvette influenced exudation by flushing a subset of cuvettes at 0·5, 4, 8, 16, 21 and 24 h. Carbon released from roots was still increasing over the first 24 h, and the increase was best described by a logarithmic function (y = 2·9205 Ln(x) + 4·3017; r2 = 0·94; Fig. 3). Thus, it seems unlikely that exudation rates were influenced to a large degree by C accumulation in the cuvette. Given the shape of the curve, shorter incubation times (e.g. 8 h) may be desirable under some conditions. However, detecting individual compounds is probably difficult with shorter incubation times particularly under conditions where exudation rates are low. Moreover, a 24 h incubation period has the added advantage in that it spans the diurnal cycle, and thus provides a time-integrated estimate that accounts for differences in exudation rates which occur between night-time and the peak hours of the day (Kuzyakov & Domanski 2000; Uselman et al. 2000).

Figure 3.

Carbon efflux from loblolly pine roots at the Duke FACTs-1 site, NC into a trap solution over the course of a week-long experiment in July, 2007. The data are best fit by a logarithmic curve (y = 2·9205 Ln(x) + 4·3017; r2 = 0·94).

We examined short-term variation of exudation by collecting exudates from the same root systems on consecutive days in July and September 2007 (Fig. 4). Such variation is important to consider given that only a limited number of root systems can be sampled on a particular day. Overall, exudation rates were positively correlated across the 2-day period (r = 0·73), and there were no significant differences between the paired data (P = 0·142). Although we cannot directly compare the responses in July and September due to the small number of samples collected, our September data suggest a trend toward reduced rates on the second day. This may have resulted from the slight reduction in average daily air temperature and PAR from day 1 to day 2 (29 °C and 926 µmol m−2 s−1 on day 1; 28 °C, 866 µmol m−2 s−1on day 2).

Figure 4.

Relationship between C efflux from eleven intact root systems over a 2 day period at the FACTS-1 site. Cuvettes were flushed five times between the days 1 and 2 collections. Open triangles were collected on 17–18 July 2007 while filled triangle were collected on 26–27 September 2007.

experiment 3

In general, exudation rates varied considerably, with the greatest variation occurring in the plot with the lowest N availability (Fig. 5). In this low N plot, exudation varied by nearly an order of magnitude – from 0·27 µg C cm−1 root day−1 in August 2007 to 2·42 µg C cm−1 root day−1 in July 2008. In contrast, exudation rates in the plot with the highest N availability varied only by a factor of c. 2 across the same time period. Overall, there were no significant differences in exudation rates among the plots (P = 0·97). Averaging across all three plots, we were also unable to detect significant differences in exudation between any two sample dates (e.g. July 2007 vs. July 2008) and among the seasons (Fig. 5 inset).

Figure 5.

Seasonal variability of mass-specific exudation from loblolly pine trees across an N availability gradient at the Duke FACTs-1 site, NC from June 2007 to July 2008. Each data point represents the mean value of two or more root systems in each plot. The inset shows the mean seasonal exudation flux from the three plots (n = 3) over the same time interval.

We also examined the degree to which root morphological variables predicted exudation (Fig. 6). We found that C efflux was well-predicted by root morphology but that such relationships depended on season. In the spring months, C efflux was predicted by the fine root surface area (r2 = 0·61; P = 0·008, Fig. 6a) and the number of root and mycorrhizal tips (r2 = 0·60; P = 0·008, Fig. 6b). Similarly, C efflux was predicted by the root surface area (r2 = 0·62; P = 0·004, Fig. 6c) and the number of tips (r2 = 0·36; P = 0·05, Fig. 6d) in the fall months. In the summer months, C efflux was somewhat correlated with surface area (r2 = 0·20) and tip number (r2 = 0·23) but these relationships were not significant (data not shown). We found no significant relationship between C efflux and root morphological variables in the winter months (data not shown).

Figure 6.

Relationship between C efflux and morphological characteristics of loblolly pine roots at the Duke FACTs-1 site, NC across three seasons. Panels on the left represent (a, c) the relationship between C efflux and fine root surface area while panels on the right (b, d) represent the relationship between C efflux and the number of fine root and mycorrhizal tips. The relationship between C efflux and pine root morphology was not statistically significant for data collected in December 2007 to March 2008.


We have demonstrated a viable new method for collecting exudates from intact root systems of mature trees in the field that overcomes many of the challenges of previous methods. Our method: (i) uses intact root systems, (ii) allows time for recovery from transplant shock, (iii) collects exudates in realistic conditions (moist glass beads rather than static solution), and (iv) can be deployed anywhere at relatively low costs. While we believe our method is a substantial step forward, linking exudation from a few individual root systems to ecosystem carbon and nutrient cycling presents some formidable challenges.

accounting for spatiotemporal variability

Accounting for the spatial and temporal variability is a persistent challenge in all root and rhizosphere studies (Hinsinger et al. 2005). Such heterogeneity presents logistical challenges in deciding how many roots need to be sampled to capture the variability of the site, and how frequently exudates need to be collected and at what time of day in order to account for temporal variation. In our analyses, we sought to reduce some of the variation in exudation by normalizing rates by fine root surface area – as exudation is predicted relatively well by this variable (Fig. 6). Still, we were unable to detect significant effects of month or season on exudation rates. Temporal variation in exudation was low at the daily/weekly scale (Fig. 4) but high over the 15-month period (Fig. 5). The high degree of variability over the course of the study was particularly evident in the plot with the lowest N availability where exudation rates varied by an order of magnitude. Because soils with low N availability often exhibit greater fine root production and mortality (Nadelhoffer et al. 1985), some of the variation in exudation may have resulted from the variable ages of roots in these plots.

Fine roots at this site live on average from 1 and 4 years (Matamala et al. 2003; Pritchard et al. 2008a; Strand 2008), and fine root production and mortality at this site occur simultaneously rather than in discrete pulses (King et al. 2002). Such differences in root life span would presumably influence exudation rates due to the physical and chemical changes that occur as fine roots age (Jones et al. 2004). An additional source of variation may be the colonization of roots by mycorrhizal fungi. Parrent et al. (2006) estimated that loblolly pine roots in these plots are colonized by 64 different ectomycorrhizal taxa, with the highest degree of phylotype richness occurring in the plots with the lowest N availability. Because mycorrhizal taxa vary in their effects on root C efflux (Ahonen-Jonnarth et al. 2000; Van Scholl et al. 2006), variability in exudation rates at the site may reflect the spatio-temporal differences in fungal taxa colonizing each root system.

Although we did not detect a significant seasonal effect, exudation rates followed the general pattern of C allocation to fine roots (Pritchard et al. 2008a) and mycorrhizal fungi (Pritchard et al. 2008b) at the FACTS-1 site, with the greatest fluxes occurring in late-spring/early summer and mid-fall. Moreover, exudation was well-predicted by the number of root and mycorrhizal tips in each cuvette. These results are consistent with reports of tips being active sites of exudation (Neumann & Romheld 2001), and suggest that mycorrhizal tips may not only be a sink for exudates but a source as well (Grayston et al. 1996). Perhaps more importantly, the results suggest that root images collected from other sources (e.g. minirhizotrons) might provide a way to derive estimates of this flux at the ecosystem-scale.

Arguably, the biggest challenge to understanding the role of exudation in forest ecosystems is not merely developing a collection method, but developing an appropriate scaling factor so that annual ecosystem fluxes (e.g. in units of g C m−2 year−1) can be calculated. A simple calculation of the annual contribution of exudates can be made by multiplying the average mass-specific exudation rate from the 15-month sampling period (10·3 × 10−5 µg C fine root biomass−1 day−1) by the average fine root biomass in the O horizon +0–15 cm of soil in these plots (c. 250 g fine root biomass m−2; R. Jackson, unpublished data). Multiplying this flux by 365 days yields an estimated annual flux of 9·4 g C m−2 year−1. This value represents 1·5% of the net primary production (NPP) in the ambient plots at FACTS-1 (625 g C m−2 year−1; from McCarthy 2006), and closely approximates estimates derived from modelling approaches (c. 1%) at this site (Luo et al. 2001). This estimate is probably conservative given that fine roots at lower soil depths, which may account for 30–50% of the total standing crop of fine root biomass at the site (Pritchard et al. 2008a), were not sampled. Thus, annual exudation rates may be closer to 3% of net primary production in this forest.

The issue of spatial and temporal variability poses a further challenge for understanding the consequences of changing exudation patterns under global change. At the Duke Forest FACTS-1 site, we have also collected exudates from trees exposed to elevated CO2 and N fertilization. Preliminary results from this experiment suggest a consistent CO2 by N interaction over several months of the growing season where CO2-induced increases in mass-specific exudation are greatest at low soil N availability (data not shown). This result is consistent with the exudation response of loblolly pine seedlings grown in a controlled environmental growth chamber (R.P. Phillips, E.S. Bernhardt & W.H. Schlesinger, unpublished) and gives us increased confidence in the potential for both independent methods to provide reliable estimates of the exudate response to each treatment. Linking the patterns and rates of exudation in seedlings in growth chambers to those of mature trees in the field may allow us to overcome the shortcomings and artefacts introduced by each method individually.

future directions

In this study, we have described a new method for collecting exudates from intact roots of loblolly pine trees, and presented data collected from over 60 root systems over a 15 month period. Given the high degree of spatial and temporal variability of this flux, we suggest that exudates will likely need to be collected over multiple years and analysed using repeated measures – similar to the approach used to understand fine root dynamics at this site (Pritchard et al. 2008a) – to understand seasonal patterns in this flux. Because the method allows for the characterization of exudate composition, realistic exudate addition experiments should also be used to examine the consequences of changing exudation patterns on soil processes (Landi et al. 2006; Kuzyakov et al. 2007; Paterson et al. 2007).

An important limitation of the method is that although root systems are left intact, the rhizosphere is not left intact due to the separation of roots and soil; this will almost certainly result in some severing of mycorrhizal fungal hyphae and may influence the flux of C from roots. Whether the latter issue can be overcome by keeping roots in the cuvettes long enough to encourage growth of extramatrical hyphae (sensu Van Scholl et al. 2006) warrants further investigation. This issue may also be resolved by coupling trap solution approaches with methods such as isotopic labelling (Hogberg 2008) and reporter gene systems (Cardon & Gage 2006) which permit the sampling of intact rhizosphere systems.


The causes and consequences of root exudation in forest ecosystems remain poorly resolved owing to difficulties in measuring rhizosphere processes. Nevertheless, several indirect lines of evidence suggest that these fluxes are probably appreciable in magnitude and highly responsive to the biotic and abiotic changes anticipated under global environmental change. Thus, a concerted effort is needed to develop novel methods for measuring this process in situ, possibly through adapting commonly used lab methods for field use.

Our method provides an important first step for measuring exudation from the intact roots of mature trees in forest soils. Although previous methods have been employed for measuring exudation rates from tree roots in situ (Smith 1976), our method offers the advantage that it is inexpensive, relatively simple to set-up, can be employed throughout the growing season, and requires only a minimal amount of soil disturbance. Moreover, our system allows for the chemical characterization of exudates, and thus should be useful in assessing the role of exudates in stimulating microbial activity, nutrient release and the decomposition of soil organic matter in forest soils. The primary caveat of the method is that the disruption of root and mycorrhizal networks present an unknown source of error to the estimates. Scaling up to ecosystem rates poses a further challenge owing to the temporal and spatial variability of exudation. Nevertheless, these uncertainties are an inevitable aspect of all in situ studies of root, rhizosphere and mycorrhizal dynamics (Read 2002).

Despite the enormous challenge, it is our opinion that a sustained effort to develop in situ methods for exudation will greatly improve our understanding of the role of below-ground processes in mediating ecosystem response to global change. However, it is important to emphasize that no one method is likely to be appropriate in all cases, and we suggest that a suite of complementary approaches (trap solutions, isotopes, reporter gene systems, microlysimetry, etc.) is likely to yield the most information on the role of rhizosphere processes in forest ecosystems.


Authors thank Elise Pendall, Lindsay Rustad and Josh Schimel for inviting us to participate in the symposium ‘Towards a Predictive Understanding of Below-ground Ecosystem Responses to Global Change’ at the 2006 SSSA meeting in Indianapolis, IN. Authors also thank Robert Nettles and David Cooley for their technical assistance at the Duke Forest FACTS1 site, Robert Jackson, Seth Pritchard, Adrien Finzi and Jeri Parrent for their insight and willingness to share unpublished data, and the members of the Bernhardt Lab for providing suggestions for improving this article. The bulk of this research was supported by the DOE FACTS-1 grant. Additional funds were provided by the Office of Science (BER), US Department of Energy, Grant No. DE-FG02-95ER62083. All meteorological data was provided courtesy of the U.S. Department of Energy Contract No. DE-AC02-98CH10886 with Brookhaven National Laboratory.