In situ high-frequency observations of mycorrhizas



  • Understanding the temporal variation of soil and root dynamics is a major step towards determining net carbon in ecosystems. We describe the installation and structure of an in situ soil observatory and sensing network consisting of an automated minirhizotron with associated soil and atmospheric sensors.
  • Ectomycorrhizal hyphae were digitized daily during 2011 in a Mediterranean climate, high-elevation coniferous forest. Hyphal length was high, but stable during winter in moist and cold soil. As soil began to warm and dry, simultaneous mortality and production indicating turnover followed precipitation events. Mortality continued through the dry season, although some hyphae persisted through the extremes. With autumn monsoons, rapid hyphal re-growth occurred following each event.
  • Relative hyphal length is dependent upon soil temperature and moisture. Soil respiration is related to the daily change in hyphal production, but not hyphal mortality.
  • Continuous sensor and observation systems can provide more accurate assessments of soil carbon dynamics.


Global climate models depend on accurate measures of carbon (C) fluxes from the atmosphere to plants and to soil. Although estimates of fluxes between the atmosphere and plant canopies have become more accurate through the use of eddy flux measurements at high temporal resolution, we have a poor understanding of the longer term allocation of net primary production (NPP) to the retention (sequestration) and respiration (Rs) of soil C (Treseder & Allen, 2000; US DOE, 2010). A better scaling approach is needed to characterize C fluxes between soils and plants.

Measurements of hyphae, mycorrhizas, and fine roots show high variation in time and space. Destructive techniques (root coring or in-growth bags) are commonly used to determine belowground NPP using a repeated harvest approach (Coleman & Crossley, 1996). Others, for example, Majdi & Nylund (1996) and Ruess et al. (2003), measured production and turnover of mycorrhizas using minirhizotrons at monthly to bi-monthly intervals. However, fine roots and fungi respond in days, not weeks to months (Allen, 1993; Stewart & Frank, 2008). Hasselquist et al. (2010) found that CO2 production was coupled to rhizomorph dynamics, but not root growth, presumably because roots are produced on very different time-scales from soil respiration. Laboratory studies of arbuscular mycorrhizal fungal hyphae estimated hyphal production and turnover on a scale of 5–7 d based either on development patterns (Friese & Allen, 1991) or 14C depletion (Staddon et al., 2003), or 9 d based on change in δ13C from in-growth bags (Godbold et al., 2006). However, Leake et al. (2004) looked at depletion and concluded that the values should be closer to 6–30 d. Meyer et al. (2010) used values of 10–100 d, from existing laboratory studies, for hyphal turnover in their modelling simulation of hyphal dynamics.

We previously described a robotic technology that measured root and fungal growth and mortality (automated minirhizotron (AMR)) placed within a networked soil sensor system consisting of measurements of temperature (T), soil moisture (θ), and soil CO2 from which we can model soil respiration (Rs) to better characterize the dynamics of hyphae, mycorrhizas and fine roots (Allen et al., 2007; Rundel et al., 2009). Here, we further describe details of the image and sensor outputs that can be used to scale up fine-root and ectomycorrhizal dynamics to provide inputs to C and nutrient models.

Materials and Methods

Our networked soil environmental observatory combines continuous monitoring using environmental sensors with daily, automated images of soil, roots and hyphae.

Site description

The research was undertaken at the James Reserve (, a California mixed-conifer forest at 1500 m with coarse textured weathered granite Entisol soil (33°48′33″N, 116°46′40″W). Soil depth ranges from 10 to 100 cm, averaging 40 cm, and the coarse roots are distributed throughout that depth, as determined using ground-penetrating radar (Stover et al., 2007). The site has a Mediterranean climate with c. 700 mm precipitation, with cool wet winters (occasional snow) and long dry summers. Additional details have been presented previously (Allen et al., 2007; Vargas & Allen, 2008a,b; Rundel et al., 2009). The study site is dominated by ectomycorrhizal plants, including ponderosa pine (Pinus ponderosa C. Lawson), sugar pine (P. lambertiana Douglas), canyon live oak (Quercus chrysolepis Liebm.), black oak (Qkelloggii Newberry), and manzanita (Arctostaphylos glandulosa Eastw.) which forms arbutoid mycorrhizas. Incense cedar (Calocedrus decurrens (Torr.) Florin), which forms arbuscular mycorrhizas (AM), is nearby, but the roots were probably not within the AMR array. Virtually all of the soil fungi sequenced in the mineral soil were known ectomycorrhizal fungi (M. F. Allen, unpublished data). All instrumentation was deployed at specific locations along the transect, ranging from locations 1 to 8, described in detail elsewhere (Vargas & Allen, 2008a,b).

Observatory instrumentation: automated minirhizotron

We utilized four AMRs for observing fine roots and mycorrhizal and saprotrophic fungi in situ (Allen et al., 2007). The AMR units are all labelled sequentially as they are produced; AMRs 5–8. These are deployed along a 30-m transect at four different locations. Four 100-mm-diameter holes were drilled using a power auger, slightly larger than the diameter of the AMR tubes. Clear glass tubes, 157 cm in length and 100 mm in diameter, were buried at a 45° angle where the soil was deep enough (> 50 cm) to accommodate the length in 2007. We then allowed the soil to repack around the tube and allowed time for fine roots and microbes to grow along the tubes. In January 2010, a robotic AMR was deployed in the tubes. The tubes are sealed with the insertion of the AMR. As the tubes are not opened for each measurement, condensation (a problem for conventional minirhizotrons) is not an issue. The AMR consists of a USB-port microscope placed on a sled that moves the camera system within the tube. The microscope camera is connected to a computer and a 120-V power source with a hot-wired connection for daily image capture. Images from these AMR units are wirelessly transmitted to a server where they are available at To describe patterns of hyphal elongation and mortality, individual images of 3.01 × 2.26 × 0.125 mm (x, y, z = depth of field) (640 × 480 pixels at ×100 magnification) were captured with the AMR. The number of images per scan per day can be preprogrammed. The maximal number of images per scan can exceed 32 000, requiring nearly 24 h to image the entire tube. For this study, we observed 4400 images taken daily using a partial scan focusing on one side of the tube.

More details on the AMR units can be found at two websites: the website of Rhizosystems, LLC which commercially produces the AMR units (, or the Center for Conservation Biology website, which contains detailed protocols, worksheets, access to all images and a description of AMR technology (

Image analysis

Images are assembled into a mosaic daily and visually evaluated for the presence of roots and fungal hyphae using RootView software, developed for the AMR ( Each image or mosaic representing a unique x, y position by date along the length of the tube was observed independently, in small mosaic units, or in movies and queried across dates to produce time-series. Each time-series is uploaded to Rootfly (Birchfield & Wells;, a software application for minirhizotron root image analysis. To measure hyphal length for each image, each hyphal branch in the initial image was digitized and the length of the line recorded and quantified by Rootfly. For each image date (daily), total hyphal length within the observable soil volume (0.850 mm3) was calculated.

For this study, five windows per tube were randomly chosen. All time sequences for these windows were observed through the study period. These 20 windows were imaged as close to daily as possible during 2011. Some days were missed when power was lost or software reboots required because of web ‘upgrades’. In addition, there were time periods, such as during the summer drought, when comparison of an image at the beginning of the period with an image at the end showed that no change had occurred. Many images captured during this drought period when there was little change are not included in this analysis.

Soil sensor network

Four sensors were deployed immediately adjacent to the four AMR units as part of a larger network of 11 sensors and conventional minirhizotron tubes installed in 2003 (Vargas & Allen, 2008a). Each location was instrumented with a sensor node including an above-ground meteorological station (HOBO Weather Station) and solid-state CO2 (Vaisala, Vantaa, Finland, CARBOCAP CO2 Transmitter Module), T (HOBO, Onset, Bourne, MA, Weather Station 12-bit Temperature Smart Sensor), and θ (HOBO Weather S-SMA Station Soil Moisture Smart Sensor) sensors at 2, 8 and 16 cm soil depths for each sensor node. The combined sensor output provides 30-min resolution basic meteorological information. Marine batteries connected to solar recharging panels supply power. The CO2 sensors are calibrated every 6 months after deployment to ensure the quality of the measurements.

Soil respiration (Rs) was calculated using a CO2 gradient flux method based on concentrations of CO2 in the soil profile. Flux-gradient Rs technology has been demonstrated using several validation applications (for details and background for this site, see Vargas & Allen, 2008a,b). The approach relies on identifying the difference in soil CO2 concentrations after applying temperature and pressure corrections at multiple depths and computing the flux based on diffusion kinetics. A gas diffusivity model is implemented to calculate soil CO2 efflux by Fick's first law where the diffusion coefficient is based on a gas tortuosity factor. Locally derived soil parameters for the model were estimated from soil physical properties and the modelled fluxes validated against manual measurements using above-ground chambers (Vargas & Allen, 2008a).

Data availability and analyses

All imagery and data sets are available at the CCB website ( Total hyphal length (mm hyphae mm−3) was measured for each image quantified. A simple, preliminary lifespan measurement was calculated for each AMR, where the lifespan was equal to standing crop (mm3)/production (mm3 d−1) of new hyphae, derived from the daily measurements.

Relative hyphal length (RHL) was calculated as the hyphal length per window proportional to the maximum hyphal length for that window. This value was the hyphal length for the individual window divided by the maximum length per window. Using daily RHL we then calculated relative daily hyphal growth rates (RHGRs) as the hyphal lengths at day 2 minus day 1, divided by day 1. Regressions were run using StatView version 5.0.


Roots, rhizomorphs, and fungal hyphae were imaged daily using the robotic AMR units (Fig. 1). Using the AMR images, we can visualize soil dynamics in response to the adjacent soil conditions and the above-ground environment at daily time-scales. Imaging is an extremely tedious step for fungal hyphae, largely because of the large amount of hyphae that is required for digitizing (Fig. 2).

Figure 1.

Mosaic and individual image from a James Reserve automated minirhizotron (AMR). The individual image, with hyphae, rhizomorphs, and soil particles is 3.01 × 2.26 mm, by 0.125 mm depth of field. Images are organized into mosaics using RootView software, developed for the AMR ( All descriptions and images are available at the Center for Conservation Biology (University of California, Riverside, CA, USA) website (

Figure 2.

A digitized image from the James Reserve from 26 August 2011. Below are images from the previous 2 d (24, 25 August), and the images for 2 d following the larger image (24, 25 August), organized into a 5 d movie to study change using RootView software ( Each hypha or other soil component is numbered for analysis of change. In this image, numbers 7 and 17 are rhizomorphs, which can be retained or subtracted, depending upon the measure-ment of interest. Fine roots can be treated identically. Digitization and measurement, for length, width, and daily change, are managed through RootFly software ( (see text for details).

Nevertheless, interesting patterns are beginning to emerge. Precipitation events such as snowfall or summer monsoonal rains increased θ so that, seasonally, T and θ were largely inversely related. Hyphal dynamics of mycorrhizal fungi were documented through the range of conditions for 2011 (Fig. 3). The peak hyphal length for the AMRs ranged from 14.47 mm mm−3 for AMR6, located between ponderosa and sugar pine and manzanita, to 29.02 mm mm−3 at AMR 8, located in the middle of an oak/pine stand. Peak mass tended to occur in the spring, when T was increasing, but θ was still relatively high. Hyphal length declined during the summer drought, but some hyphae remained. Regrowth was rapid with the late summer monsoonal precipitation events, with a large jump in hyphal mass following the large rainfall in early October.

Figure 3.

Relative length of hyphae from the study transect at the James Reserve through 2011. Shown are individual relative hyphal lengths for the four automated minirhizotron (AMR) units showing the variation within a stand. Also shown are the changing average temperature (T), soil moisture (θ) and soil respiration (Rs) through the year (see text for details).

Individual hyphae are highly dynamic but simultaneously resilient. Some individual hyphae were observed growing as much as 800 μm d−1 and an individual RHGR as high as 2.4 was observed. Lifespan estimates are highly variable both among AMR units and even among observation individual windows. The hyphal lifespans ranged from 117 d at location 6 to 33 d at location 3. The overall mean lifespan for all windows was 48 d.

Hyphae survived for long periods without change, but then changed rapidly. One ectomycorrhiza and the extending hyphal network (location = 5, = 18 of AMR5) persisted during cool, wet conditions (Fig. 4), persisting virtually intact until after 11 March, when the external hyphae virtually disappeared. The hyphae regrew, died and regrew. The hyphae emanating from this particular ectomycorrhiza were then replaced as new hyphae regrew from a neighbouring ectomycorrhiza. The individual ectomycorrhiza of Fig. 4 persisted until 12 January 2012, and then died and decayed. As it decomposed, a neighbouring ectomycorrhiza recolonized by 24 January 2012.

Figure 4.

An ectomycorrhiza (EM) and radiating hyphae under snow, from 21 January 2011, automated minirhizotron 5 (AMR5). Shown (clockwise from right) is a portion of the mosaic (a), with one EM highlighted by the arrow. A higher resolution mosaic (b) shows nine linked images, and (c) a single image from the middle of the upper row of (b). Each image (square) is 3.01 × 2.26 mm, by 0.125 mm depth of field.

As exemplified by the window described in the previous paragraph, we saw that the hyphae were relatively stable through winter with cool temperatures and high soil moisture. In late January, following a dry period, there was some hyphal mortality, but overall hyphal lengths only changed slightly. By late March, as soil dried out, there was a simultaneous period of high mortality and high production of new hyphae. This corresponded to a short drought event in early March, followed by a snowfall and melting in late March (Fig. 5). The ectomycorrhiza observed in Fig. 4 disappeared, and new ones nearby were formed. While the total hyphal length showed some change, the overall length change was not dramatic (Fig. 5). Hyphae rapidly grew with the autumn rains (Fig. 3) at all locations.

Figure 5.

Daily production and mortality of hyphae in the winter and spring of 2011, along with the overall hyphal length in response to daily averaged soil temperature at 8 cm, soil moisture (θ) at 8 cm, and soil respiration (Rs), expressed as soil CO2 efflux. The observations are all from a single point, automated minirhizotron 5 (AMR5).

To scale up to the stand level, we evaluated the response of relative hyphal length to changing T and θ. Overall, relative hyphal length was negatively related to T and positively related to θ. Using multiple regression,

display math

Hyphal length varied among AMR units but RHL shifted rapidly in all units with changes in θ and T. One location (AMR6, located between oak, pine and manzanita) sustained high lengths of hyphae during the dry season. The other locations had very low hyphal activity during peak drought. However, there was massive mortality with the first fall monsoon followed by rapid regrowth.

By taking and quantifying daily images, we calculated the RHGRs. No significant relationships were found by regressing RHGR with T, θ, or CO2. However, by separately examining production (positive hyphal growth) and mortality (negative hyphal growth), patterns emerged. The RHGRs for production and mortality were both related to T and θ, where:

display math


display math

Examining T and θ separately, production was low at freezing temperatures, and increased (slope of +0.005; = 0.018) up to 17.5–20°C. Above 20°C, θ rapidly declined (Fig. 2) and new production was nearly 0, such that RHGRproduction declined (slope of −0.565; = 0.03). Mortality showed the opposite pattern, in that as T increased from freezing, mortality (or negative growth) continued to increase (became more negative; slope = −0.03; < 0.0001). The greatest mortality occurred at low θ (slope = 0.181; = 0.007), when T was high.

Both hyphal production and mortality occurred at the entire range of T, but peaked at intermediate T. Specifically, the hyphal production increased as soil T rose from near freezing to intermediate T values. Thus, the regression was actually constrained and inappropriate at higher T. The greatest mortality (negative growth) occurred at intermediate T. At low T and high θ, hyphae tended to persist but not grow, whereas at high T and low θ, mortality tended to be high. During the extremely dry periods (that had high T ), there was little production but continued mortality.

RHGR(production) growth occurred preferentially at high θ. New hyphae were produced optimally at between 18% and 23% θ. This rapid hyphal growth phase occurred in the late fall, with late monsoonal rains, when T was still high but had begun to decline.

Rs was related to RHGR, but in a complex way. When we looked just at production, Rs was significantly correlated (= 0.13; = 0.02) to RHGRproduction. While this relationship explained only a fraction of the variation in Rs, it was nevertheless statistically significant, where:

display math

RHGRmortality was not significantly related to Rs.


Mycorrhizas make a critical contribution to soil C and Rs. Between 10 and 35% of net primary production (NPP) is allocated from plants to mycorrhizal fungi (Finlay & Soderstrom, 1992; Treseder & Allen, 2000; Allen et al., 2003) as well as forming recalcitrant C compounds (Treseder & Allen, 2000). These fungi form a critical C bridge between plant roots and soils. By better documenting hyphal growth and mortality, we should be able to develop more accurate measures of nutrient and C fluxes within soils.

Our observations showed that hyphal dynamics were coupled to the Mediterranean-type climate of southern California. Overall, hyphae grew and died constantly, and production and mortality were overlapping. During the cool wet periods, partially under snow, hyphae persisted with little change. As dry periods started in spring, with each snow or rain event, mortality was high, but there was simultaneous regrowth. During this period, there was minimal overall change in hyphal length or mass, but a high turnover. In simply measuring standing crop from cores or infrequent observation, NPP would be underestimated. These conditions also represent an opportunity for assessing turnover in species composition. Preliminary sequence analyses from cores indicate 92% fungal species turnover within 1 wk at a forest edge c. 1 km away, during the spring, compared with 75% during the dry season (M.F. Allen, unpublished data). However, even using adjacent cores, we could not separate temporal from spatial turnover. As the soils continued to dry down, there was pronounced hyphal mortality, exceeding new hyphal production. At AMR6, hyphae persisted through the dry period. We believe that this persistence was tied to observed hydraulic lift (Querejeta et al., 2003, 2009; K. Kitajima, M.F. Allen & M. Goulden, unpublished data). Autumn monsoonal precipitation events resulted in new hyphal growth. The primary growth spurt occurred with each late fall monsoonal precipitation as T declined.

The average hyphal lifespan was 48 d (range 33–117 d among the AMR units), longer than we expected, and longer than most of the estimates of 5–9 d (Friese & Allen, 1991; Staddon et al., 2003; Godbold et al., 2006), and more similar to the projected turnover estimates of 30 d by Leake et al. (2004). Hyphal lifespans were still far shorter than the lifespans of both rhizomorphs and fine roots at the same site (Kitajima et al., 2010). Although Ruess et al. (2003) found that birth and death of mycorrhizal roots was related to season, in this ecosystem we found that production of new hyphae and their mortality were simultaneous and tied to precipitation and drought events. Individual hyphae grew rapidly, but then often persisted through several events. During the same period, other hyphae produced earlier died, and new hyphae were produced. The wide variation among point locations makes averaging for scaling difficult. However, our estimate of an average lifespan of 48 d means an annual turnover of 7.6 or a daily turnover of 2.1%, well within the range modelled (Meyer et al., 2010). Nevertheless, the finding of a range from 33 to 111 d for individual locations means that more detailed understanding of the factors involved at each point, and at more points at a single location, is needed to accurately model stand-level C allocations. Our next step is to integrate the observations of individual hyphae with mark–recapture models to assess hyphal turnover, as has been undertaken for fine roots (Ruess et al., 2003; Kitajima et al., 2010). As we begin to identify individual hyphae and differentiate morphological types more detailed modelling should provide much more useful information on changing mycorrhizal C allocation with environmental change.

The relationship between Rs and hyphal dynamics was complex. Rs increased as hyphal production increased. Although RHGR explained only a small part of the variation (r2 = 0.017) in changing Rs, it was nevertheless coincidental. Thus, when the mycorrhizal hyphae are growing more rapidly, roots and other soil organisms probably increase their activity, as measured by soil respiration. There was no relationship between Rs and RHGRmortality. The respiration of C from hyphal mortality is probably a component of a larger process of fine-root mortality and decomposition. At this point, we have not teased apart metabolic, growth and decomposition respiration, nor the different groups of soil organisms involved. However, these AMR measurements coupled to analysis of new versus old C (δ14C or δ13C) in the respired CO2 hold promise. In addition, there are distinct lags in Rs that we need to understand better to incorporate into descriptions of the dynamics of hyphae.

Production and mortality of hyphae are presumed to be rapid and highly variable, but we know little about how they actually behave in the field. In the laboratory, AM fungi are produced and disappear within days (Friese & Allen, 1991; Staddon et al., 2003). Ectomycorrhizal root tips and rhizomorphs (Treseder et al., 2005) have variable lifespans (from days to a decade) (Ruess et al., 2003; Hasselquist et al., 2010). Rs is sensitive to rapid (hourly/daily) changes in growth and mortality of individual hyphae (Vargas & Allen, 2008b). At each critical event, such as a monsoonal rain or a snowstorm, a variable fraction of the hyphae die and new ones are born. In analysis of microbial mass through critical time periods such as spring growth (observational, biochemical, or indirect analysis), data from soil cores show only change in mass and cannot identify turnover. Thus, NPP is underestimated. However, by tracking individual hyphae, we can determine C allocations to hyphal production.

In trying to examine how differing ecosystems will respond to climate change, we need to better understand the fine-scale drivers of soil ecosystem dynamics, including seasonality of production and species taxonomic turnover. Soils are a large sink globally of C and are increasingly recognized as dynamic simultaneous sources and sinks of C, depending on the temperature, moisture, and vegetation cover of a site. More and better coupling of soil observatories and sensors open the ‘black-box’, particularly by better understanding production and turnover, crucial to the scaling of soil processes to large-scale ecosystem models, and will provide better resolution of source/sink strength and more accurate ecosystem modelling.


We thank Rebecca Hernandez, Hai Vo, Renee Wang, Tim Mok, Stephanie Ngo, Kevin Choy, and Daniel Tran who helped us with processing image data, and Mike Taggart and Tom Unwin at Rhizosystems Inc. who developed and maintained the automated minirhizotrons. Special thanks go to Michael Hamilton and Becca Fenwick at the James Mt Jacinto Nature Reserve for the support during study. This research was funded by the National Science Foundation (EF-0410408 and CRR-0120778).