Root system architecture (RSA) encompasses structural features that provide spatial configuration such as root length, width, spread and number (Khan et al., 2016; Figure 1) and is an important rhizosphere parameter in regulating soil porosity, and nutrient and water uptake efficiency by plants (Helliwell et al., 2017; Fang et al., 2019). Plants have been observed to “sense” and direct root growth toward nutrient sources in soil, and the RSA of a plant exhibits great malleability in response to environmental stimuli (changes in nutrients, pH, soil moisture, and temperature) which in turn, influences microbial communities (Bao et al., 2014; Saleem et al., 2018). For instance, bean plants grew deeper roots under drought conditions to enhance water foraging capabilities while low phosphate (P) conditions stimulated the formation of dense lateral roots involved in P uptake from upper soil layers (Ho et al., 2005). Given that most soils are heterogenous, understanding the RSA of plants becomes critical in improving resource use efficiency and agricultural yields (Ingram and Malamy, 2010; Khan et al., 2016). Often, RSA in pot-grown plants is investigated by excising the roots via mechanical means such as root washing or blowing with compressed air (Judd et al., 2015). These methods are, however, time-consuming, cause inevitable damage of fine root hairs and result in loss of spatial and temporal information (Judd et al., 2015).

An appealing alternative for studying RSA is the use of rhizotrons. Rhizotrons were initially constructed as underground facilities designed for viewing and measuring roots in the field (Klepper and Kaspar, 1994). In the lab, the rhizotron implies a chamber constructed using two vertical sheets with at least one or both of the sheets being transparent and/or removable (Figure 2A). This allows repeated visual inspections of individual roots; a feature unachievable with destructive sampling. In some cases, the word “rhizobox” is used for a similar set up although this was first introduced in as compartmentalized systems to separate the root and soil compartments (Kuchenbuch and Jungk, 1982). Rhizotrons/rhizoboxes are often constructed with PVC or acrylic materials and come in many sizes to accommodate different plants with soil or soil-less substrates (Neumann et al., 2009). Root growth and morphology in the rhizotron can be tracked by a variety of methods ranging from manual tracing onto a plastic sheet, using handheld or flatbed scanners to fully automated time-lapse imaging camera systems (Mohamed et al., 2017). Data can be subsequently analyzed with a wide range of software packages (Kuijken et al., 2015). Affordable and robust RSA imaging platforms using rhizotrons have also been developed for increased accessibility in low-income countries (Bontpart et al., 2020).

The versatile construction of a rhizotron design for RSA studies has inspired many variations. For instance, ara-rhizotrons were designed to enable the study of 3D canopy competition with simultaneous root growth observation in an Arabidopsis plant population (Devienne-Barret et al., 2006). The horizontal and radial design of Horhizotron^TM and mini-Horhizotron consisting of transparent quadrants attached to a central chamber were developed to study lateral growth of roots in a semi-3D space and to perform post-transplant assessment (Wright and Wright, 2004; Judd et al., 2014). The separated quadrants can also be used with different soil substrates simultaneously to study substrate effects on root growth (Wright and Wright, 2004). A rhizotron fitted with water-tight gasket seals has also been used successfully to investigate the RSA of plants under water-logged conditions (Busch et al., 2006).

Despite the continuous real-time visual read-out, most rhizotron designs suffer from inevitable loss of information from roots occluded by soil particles. The GLO-Roots system overcomes this by imaging from both sides of the rhizotron while using bioluminescent roots to create higher contrast against the soil, enabling quantitative studies on RSA (Rellán-Álvarez et al., 2015). Following advances in engineering and device fabrication, more rhizotron variants adapted to specific plant growth conditions can be envisioned.

In a typical topsoil, approximately half is composed of solid minerals and organic matter while the rest is a fluctuating composition of water and gas filled spaces influenced by environmental conditions and uptake/release of solutes from plants (O’Donnell et al., 2007). Changes in gaseous and hydrologic parameters, such as ions, O₂ and moisture among others, create a spatially complex environment that influences microbial communities and overall plant health. These physicochemical fluxes are heterogeneously distributed along roots and vary with root types and zones (Neumann et al., 2009). Often, they exist as gradients in the rhizosphere (Kuzyakov and Razavi, 2019), thus emphasizing the need for non-destructive sampling in order to accurately capture processes occurring at biologically relevant times and scales.

Rhizotron chambers with a visually accessible rhizosphere allows in situ and continuous mapping of these gradients in the soil through the use of different types of imaging methods. For instance, photoluminescence-based optical sensors enable in situ, repeated detection of small molecule analytes in addition to pH (Blossfeld et al., 2010), O₂ (Frederiksen and Glud, 2006) and NH₄ (Santner et al., 2015). Methods like zymography to detect enzyme activity (Spohn et al., 2013) and diffusive gradients in thin film (Santner et al., 2012; Valentinuzzi et al., 2015) can be used to map solute concentrations in the soil down to sub-mm scales with high spatial resolution more realistically than traditional destructive approaches. For example, transport and distribution of water in the rhizosphere soil has been imaged on both 2D and 3D planes by coupling a rhizotron with neutron radiography and tomography, respectively (Esser et al., 2010) and showed varying moisture gradients along the root system with higher water uptake at the rhizosphere compared to bulk soil. On the other hand, if the rhizotron slabs are thin enough (∼4 mm), even simple imaging solutions based on light transmission can be set up to capture water uptake by roots in sand (Garrigues et al., 2006). Despite trade-offs in method sensitivity between these two studies, a rhizotron set up is critical in both designs and illustrates its adaptability to multiple equipment.

Roots exude a substantial amount of photosynthetically fixed organic carbon into the soil consisting of a wide variety of compounds such as sugars, organic acids, and primary and secondary metabolites (Sasse et al., 2018; de la Fuente Cantó et al., 2020). Together with mucilage and border cells (which are mainly expelled from root tips), root exudates provide a major source of nutrients for the rhizosphere microbiome (Figure 1). Root exudation is regulated under genetic control (i.e., genotype, root type and developmental stage) (Canarini et al., 2019) as well as in response to environmental conditions in the soil such as nutrient limitations or increase in toxicity (van Dam and Bouwmeester, 2016). Exudate patterns are also recognized as one of the strongest drivers shaping the rhizosphere microbiome (Dessaux et al., 2016; Zhalnina et al., 2018b; de la Fuente Cantó et al., 2020). As a central player in the rhizosphere ecosystem, it is imperative to understand root exudation patterns to unravel subsequent impacts to the surrounding soil and microbial community.

Improvements in analytical instrumentation have made it possible to move from targeted to untargeted explorations with mass spectrometry to create root exudate fingerprints in its entire complexity (Oburger and Schmidt, 2016). Regardless, the impact of such techniques relies partly on our exudate sampling techniques. Detection of exudates in real-time is difficult due to rapid biotransformation and sorption to the soil matrix. As such, common collection methods rely on root washing in hydroponic systems to overcome complications in the soil matrix and preserve native exudation profiles. However, a comparison between a soil-based collection method and hydroponic methods showed varied responses particularly in amino acid exudation although the underlying cause was not elucidated (Oburger et al., 2013). It is possible that the differing growth conditions between hydroponics and soil, which include differences in gas concentrations, mechanical impedance and microbial spatial composition, can elicit differing root exudation responses to the same environmental stimuli.

Rhizoboxes offer the advantage of localized sampling in soil using sorption media such as paper and membrane filters, compound specific ion exchange binding resin or micro-suction cups placed closed to root zones of interest to collect exudates (Kamh et al., 1999; Neumann et al., 2009, 2014). Moreover, in a rhizobox fitted at the bottom with a porous root-impenetrable membrane, a root mat is allowed to be formed which is then further transferred onto a collection compartment (Figure 2D; Chaignon et al., 2002; Chaignon and Hinsinger, 2003). The collection compartment containing soil could then be cut into thin slices (1–3 mm) parallel to the membrane to represent differing distances from the rhizosphere (Neumann et al., 2009). While this approach can be used to investigate exudate release and sorption under soil conditions, the root mat growth generalizes exudate production in terms of the whole root system and occludes spatial exudation patterns. In a hybrid set up by Oburger et al. (2013), the rhizobox is transplanted to a second specialized rhizobox for continued vertical root growth. This specialized rhizobox consists of a nylon membrane (30 μm pore size) close to the transparent side to restrict root growth into the soil except for root hairs (Figure 2C). This creates a vertical flat root mat onto which localized exudate samples can be collected. A comparison of this novel set up to conventional collection methods showed that amino acid exudation rates were most varied among the different methods (Oburger et al., 2013), further highlighting the need for specialized chambers.

Nonetheless, successful implementation of these chambers is still limited to fast-growing plants which can form active root mats. The high density of root mats could also lead to unnatural root exudate levels and an overestimation of rhizosphere effects. In addition, care has to be given to the choice of membrane as selective sorption of certain root exudates onto the membrane may also occur (Neumann et al., 2009).

Free-living nematodes are ubiquitous in the soil. They are beneficial to the plants by playing a role in nutrient cycling and in defense against insects and microbial infections through signaling interactions with the roots (Rasmann et al., 2005; Manosalva et al., 2015; Figure 1). Conversely, infections by parasitic nematodes in the roots increase the plant’s susceptibility to stress and other pathogenic bacteria, fungi, and viruses creating major losses in crop productivity (Powell, 1971; O’Callaghan et al., 2018). With an impending rise in nematode infections due to climate change, understanding nematode behavior and interactions in the rhizosphere becomes important to develop appropriate biocontrol methods to ensure long term food security (O’Callaghan et al., 2018).

Traditional nematode studies are performed in petri dishes with agar or culture media (Dinh et al., 2014; O’Callaghan et al., 2018). However, these substrates do not accurately emulate the physical textures and heterogeneity of soil and create homogenous solute and temperature gradients which could impact nematode behavior and interactions with the roots (Lockery et al., 2008). Indeed, nematode motility speed and dispersal decreased in substrates more closely mimicking sand (Hapca et al., 2007). On the other hand, studying nematode behavior in the soil is a difficult endeavor as its near-transparent body and small size makes it almost indistinguishable from soil particles. Cross-sectioning and staining infected roots make it possible for nematode visualization but they are destructive and provide only static snapshots of cellular changes or nematode behavior during infections (Dinh et al., 2014).

On the other hand, microscopy rhizosphere chambers provide non-invasive detection and observation of nematode activity in the rhizosphere (Froelich et al., 2011; Kooliyottil et al., 2017). The roots in these chambers grow between a glass slide and a nylon membrane (unknown pore size) (Figure 2C). The membrane restricts movement of roots except root hairs into the soil while the transparent glass enables microscopy of the roots at high resolution (Froelich et al., 2011). Coupled with fluorescently stained nematodes, microscopy rhizosphere chambers allowed for non-destructive in situ observations of nematode infection in its host species over the entire life of the parasite (Dinh et al., 2014; Kooliyottil et al., 2016).

Nonetheless, staining nematodes is an additional challenge as nematode cuticles are impermeable to stains (O’Callaghan et al., 2018). This can, however, be alleviated by using advanced imaging technologies which eliminates the need for staining. A recent study demonstrated live screening of nematode-root interactions in a transparent soil-like substrate through the use of label-free light sheet imaging termed Biospeckle Selective Plane Illumination Microscopy (BSPIM) coupled with Confocal Laser Scanning Microscopy (Downie et al., 2014; O’Callaghan et al., 2018). Using this set up, researchers were able to monitor roots for nematode activity at high resolution and suggest its possible use in rapid testing of chemical control agents against parasitic nematodes in soil-like conditions (O’Callaghan et al., 2018).

Fungal communities in the rhizosphere are involved in the degradation of organic matter in the soil and subsequent nutrient turnover affecting plant health as well as the microbial community (Buée et al., 2009). Fungal biomass often reaches a third of total microbial biomass carbon (Joergensen, 2000) and almost all terrestrial plants are able to form symbiotic associations with mycorrhizal fungi (Van Der Heijden et al., 2016; Figure 1). The majority of these associations are with arbuscular mycorrhiza fungi (AMF) (Smith and Read, 2008) which penetrate into root cortex cells to form highly branched structures (Harrison, 2005). The investment of photosynthetic carbon by plants to AMF is rewarded with increased nutrient availability made possible by the extended hyphal network in the soil. For instance, up to 90% of phosphorus uptake in plants can be contributed by symbiosis with AMF (Ferrol et al., 2019). AMF networks in the soil also influence water retention and soil aggregation further impacting plant growth (Augé, 2004). Moreover, next-generation sequencing technologies and advances in imaging techniques have greatly improved our knowledge on the taxonomical and functional properties of fungal communities in the rhizosphere (Oburger and Schmidt, 2016). However, these methods are optimized for fine scale (millimeter) analysis and are not capable of assessing the foraging capabilities of hyphal networks which can span across centimeter to meter scales.

Toward this end, several researchers have used compartment setups with physical barriers created by 20–37 μm nylon membranes (Figure 2E) which restrict movement of roots but not mycorrhizal fungi. This separation creates root-free and plant-free soil compartments connected only by mycorrhizal fungi to examine the transport of various compounds across these compartments. Using this set up, the importance of mycorrhizal fungi in the flow of different elements such as carbon (Kaiser et al., 2015), nitrogen (Tanaka and Yano, 2005) and phosphorus (Wang et al., 2016) between plants, soil and microbes over centimeter distances have been validated. Repeated disruption of the hyphal connections also led to a decreased resistance in plants to drought stress (Zou et al., 2015). The membranes can also be placed horizontally to create different depth gradients to investigate hyphal contributions to water uptake (Ruiz-Lozano and Azcón, 1995). In some studies, an additional 1.5–3 mm air gap is created between two membranes with a wire net to restrict solute movement between two chambers (Tanaka and Yano, 2005; Zhang et al., 2010; Koegel et al., 2013; Figure 2E). A common feature of these set ups is the size-exclusion membranes which proved to be critical in distinguishing fungal hyphae processes in the rhizosphere soil.

In addition to AMF interactions, a split root set up, which separates the roots of one plant into halves, can be introduced to investigate the systemic response of plants (Figure 2F; Vierheilig et al., 2000). In essence, the split-root system directs the growth of the roots to generally two different growth conditions and enables the investigation of whether a local stimuli (microbial interactions, nutrient limitations, etc.) have a local or global response which can be observed at the root or shoot level (Agapit et al., 2020). Split-root systems are widely studied (Larrainzar et al., 2014; Saiz-Fernández et al., 2021) and have been adapted to rhizoboxes (Zhu and Yao, 2004; Mitchell et al., 2018) as well as to pots and tubes (Kosslak and Bohlool, 1984; Marschner and Baumann, 2003).

In the rhizosphere, plants host a wide diversity of bacteria on the surface of the root (epiphytes) as well as within roots in the vascular tissue (endophytes). Due to its abundance and importance, the bacterial community in the rhizosphere is perhaps the most widely studied among other microbial members in the rhizosphere ecosystem. While the study of endophytic bacteria requires inevitable destructive sampling due to its localization, several non-destructive approaches have been developed to study microbes inhabiting the rhizoplane.

One of the most widely studied plant-microbe interactions in the rhizosphere is that of the symbiotic relationship between legumes and rhizobia (Hirsch et al., 2001). Once a potential nodule forming bacteria is isolated, it is often required to authenticate its nodule forming phenotype by inoculating on host plants. However, conventional methods such as the use of soil pouches do not allow long term incubation, while “Leonard jars,” consisting of two stacked glass jars forming the top soil layer and the bottom nutrient solution layer, can be expensive and time consuming (Yates et al., 2016). A recent study challenges this by describing the use of clear plastic CD cases as mini-rhizotrons with potential for use in phenotyping root traits such as legume formation, and demonstrated innovation that democratizes research opportunities in rhizosphere research (Cassidy et al., 2020).

Other microbial interactions in the rhizosphere, however, may not result in visible changes to the root system and often rely on next-generation omics technologies. As such, physical separation of the rhizosphere from the bulk soil becomes paramount in elucidating changes to microbial community and interactions. One approach to this end is the use of nylon bags with differing pore sizes (10–50 μm) (Figure 2G). The nylon bag restricts the movement of roots and the soil inside the bag is then regarded as the rhizosphere soil to compare against the surrounding root-free bulk soil (Yevdokimov et al., 2006; Shrestha et al., 2010; Nie et al., 2015). Developing further on this concept, Wei et al. (2019) designed a specialized rhizobox that allowed repeated non-destructive sampling by adding individual nylon bags of root-free soil surrounding the root compartment which are then used as a proxy for the rhizosphere (Wei et al., 2019).

These methods allowed easy distinction of the rhizosphere and the bulk soil but, we now know that the rhizosphere community is not only distinct from the bulk soil but also varies with type, part and age of the root, largely as a consequence of varying root exudation patterns (Sasse et al., 2018). Studying this phenomenon in situ in the soil requires separation of desired roots from others without disturbance to plant growth or soil. To address this, researchers have used a modified rhizobox design with a side compartment to regulate root growth and quarantine specific roots from the main plant chamber (Figure 2B). This additionally creates easy distinction between old and new roots and allows testing on specific quarantined roots despite plant age. A study using this set up showed specific microbial chemotaxis toward different exudates (sucrose or tryptophan) on an individual root (Jaeger et al., 1999) whereas another showed spatial and temporal regulation of niche differentiation in microbial rhizosphere guilds (Nuccio et al., 2020). Similar physical perturbations to regulate root growth in response to microbial stimuli have also been applied in the microscale and are explored in the next section.

A complex web of biochemical processes and interactions occur in microscale dimensions in the rhizosphere. Having the ability to interrogate and manipulate these microscale processes and environmental conditions with high spatiotemporal resolution will elucidate mechanistic understanding of the processes. Microfluidics has proven to be a powerful approach to minimize reagent usage and to automate the often-repetitive steps. The microscale of the channels also allows precise control of reproducible conditions utilizing the laminar flow and automated fluidic operations (Figures 3A,B). In addition, the microfluidic devices are well integrated with conventional imaging techniques by using a glass slide or coverslip as a substrate bonded with polydimethylsiloxane (PDMS). These characteristics, as well as the ability to rapidly prototype and reproducibly manufacture using soft lithography technique, have enabled new ways of interrogating and studying the rhizosphere environment in a reproducible manner.

Many of the microfluidic devices used for studying the rhizosphere share a similar design concept (Khan et al., 2019). They have an opening port, sometimes with pipette tips inserted into the PDMS body where the seed of the seedling rests and a microchannel where the primary root grows into. The dimension of the channel depends on the type and age of the plant. For example, an Arabidopsis thaliana’s seedling is typically grown in a microfluidic device up to 10 days, with chamber dimension around 150 to 200 μm in height, whereas the Brachypodium distachyon seedling chamber is 1 mm in height due to its thicker roots (Massalha et al., 2017; Khan et al., 2019). Media and/or inoculation of the microbiome is achieved through additional channels to the main chamber. The PDMS body with the channels is typically bonded on a 50 mm by 75 mm microscope slide, and is made to accommodate multiple plants to increase throughput. Automated control offers the ability for continuous imaging and manipulation of media conditions with high temporal resolution.

One notable example of a microfluidic device for rhizosphere studies is the RootChip, which uses the micro-valves in a PDMS device to control the fluidics (Unger et al., 2000; Grossmann et al., 2011). The first study using the RootChip grew 8 Arabidopsis plants on a single device with micro-valves (Grossmann et al., 2011) but by the second iteration, the throughput has been doubled (Jones et al., 2014; Keinath et al., 2015) indicating rapid technological advances in the field. In addition, all these studies demonstrated spatiotemporal imaging at single-cell resolution and dynamic control of the abiotic environments in the rhizosphere.

Another microfluidics-specific application to rhizosphere study is to use the laminar flow to generate the spatially precise and distinct microenvironment to a section of the root as demonstrated by Meier et al. (2010). A young Arabidopsis’ seedling was sandwiched and clamped between two layers of PDMS slabs with microchannel features to tightly control synthetic plant hormone flow with 10 to 800 μm resolution to the root tip area, enabling observations of root tissues’ response to the hormones. As many root bacteria produce auxin to stimulate the interactions with the root, this study showed the possible mechanism of microbiome inducing the interaction by stimulating root hair growth. Another application of laminar flow utilized the RootChip architecture by adding the two flanking input channels to generate two co-laminar flows in the root chamber, subjecting a root to two different environmental conditions along the axial direction to study root cells adaptation to the microenvironment at a local level (Stanley et al., 2018). These studies revealed locally asymmetrical growth and gene pattern regulations in Arabidopsis root in response to different environmental stimuli.

Microfluidic platforms have also been successfully employed to study the interactions between the root, microbiome and nematodes in real time (Parashar and Pandey, 2011; Massalha et al., 2017; Aufrecht et al., 2018). In the systems, additional vertical side channels are connected perpendicularly to the main microchannel to enable introduction of microorganisms and solutes to the roots in a spatially and temporally defined manner (Parashar and Pandey, 2011; Aufrecht et al., 2018). A recent microfluidic design incorporated a nano-porous interface which confines the root in place while enabling metabolite sampling from different parts of the root (Patabadige et al., 2019). These studies demonstrated the potential of microfluidics in achieving spatiotemporal insights into the complex interaction networks in the rhizosphere.

Despite several advantages of microfluidics in rhizosphere research as described above, some challenges remain. All the microfluidic applications grow plants in hydroponic systems where clear media is necessary for the imaging applications and packing solid substrates in the micro-channels is not trivial. The microscale of the channels limits the applications of these devices to young seedlings. Thus, interrogating the microscale interactions in bigger, more developed plants is not possible with current microfluidic channel configurations. In addition, technical challenges such as operating the micro-valves and microfabrication present a barrier to device design and construction for non-specialists.

Fabricated ecosystems aim to capture critical aspects of ecosystem dynamics within highly controlled laboratory environments (Zengler et al., 2019). They hold promise in accelerating the translation of lab-based studies to field applications and advance science from correlative and observational insights to mechanistic understanding. Pilot scale enclosed ecosystem chambers such as EcoPODs, EcoTrons and EcoCELLs have been developed for such a purpose (Griffin et al., 1996; Lawton, 1996; Ke et al., 2020). These state-of-the-art systems offer the ability to manipulate many parameters such as temperature, humidity, gas composition, etc., to mimic field conditions and are equipped with multiple analytical instruments to link below ground rhizosphere processes to above ground observations and vice versa (Figure 3C; Griffin et al., 1996; Lawton, 1996; Ke et al., 2020). Currently, however, accessibility to such systems is low as there are only several places in the world which can host such multifaceted facilities due to the requirement of significant financial investments.

Switching back to lab-scale systems, a recent perspective paper calls for the need to standardize devices, microbiomes and laboratory techniques to create model ecosystems (Zengler et al., 2019) to enable elucidation of molecular mechanisms mediating observed plant-microbe interactions e.g., exudate driven bacterial recruitment (Zhalnina et al., 2018a, b). Toward this goal, open source 3D printable chambers, termed Ecosystem Fabrication (EcoFAB) devices, have been released with detailed protocols to provide controlled laboratory habitats aimed at promoting mechanistic studies of plant-microbe interactions (Gao et al., 2018). Similar to a rhizotron setup, these flow-through systems are designed to provide clear visual access to the rhizosphere with flexibility of use with either soil or liquid substrates (Figures 3D,E). Certainly, there are many limitations to these devices (discussed in more in Table 1) in that they are limited to relatively small plants and limit the 3D architecture of the root system. Still, an advantage with the EcoFAB is that its 3D printable nature allows for adaptations and modifications to be made and shared on public data platforms such as Github for ease of standardization across different labs and experiments (Sasse et al., 2019). In fact, a recent multi-lab effort showed high reproducibility of root physiological and morphological traits in EcoFAB-grown Brachypodium distachyon plants (Sasse et al., 2019). The development of comparable datasets through the use of standardized systems is crucial to advancing our understanding of complex rhizosphere interactions. Open science programs such as the EcoFAB foster a transparent and collaborative network in an increasingly multi-disciplinary scientific community.