Soil conditions and cereal root system architecture: review and considerations for linking Darwin and Weaver

Charles Darwin founded root system architecture research in 1880 when he described a root bending with gravity. Curving, elongating, and branching are the three cellular processes in roots that underlie root architecture. Together they determine the distribution of roots through soil and time, and hence the plants’ access to water and nutrients, and anchorage. Most knowledge of these cellular processes comes from seedlings of the model dicotyledon, Arabidopsis , grown in soil-less conditions with single treatments. Root systems in the field, however, face multiple stimuli that interact with the plant genetics to result in the root system architecture. Here we review how soil conditions influence root system architecture; focusing on cereals. Cereals provide half of human calories, and their root systems differ from those of dicotyledons. We find that few controlled-environment studies combine more than one soil stimulus and, those that do, highlight the complexity of responses. Most studies are conducted on seedling roots; those on adult roots generally show low correlations to seedling studies. Few field studies report root and soil conditions. Until technologies are available to track root architecture in the field, soil analyses combined with knowledge of the effects of factors on elongation and gravitropism could be ranked to better predict the interaction between genetics and environment (G×E) for a given crop. Understanding how soil conditions regulate root architecture can be effectively used to design soil management and plant genetics that best exploit synergies from G×E of roots.


Darwin and root architecture research
In 1880, Darwin elegantly described how a single root tip bends towards gravity. A century and a half later, the molecular processes underlying the events in the root tip during gravitropism are documented. Gravitropism is one of the most studied root processes, and has been elucidated primarily through studies in soil-free conditions that impose a gravitational treatment on young plants. In the 1920s, Weaver's meticulous drawings of root systems down soil profiles illustrated clearly that the sensing of gravity is a key determinant of architecture (Weaver et al., 1922;Weaver, 1925). Weaver's drawings highlight that adult root system architecture depends on species, soil conditions, and root type. Recent reviews (Jovanovic et al., 2007;Osmont et al., 2007;Hodge et al., 2009;Ingram and Malamy, 2010) address the influence of factors on processes underlying root architecture. However, most data come from experiments where treatments are applied singly in controlled conditions, with an emphasis on molecular studies with the model dicotyledonous plant Arabidopsis. Root systems start as a single root formed during embryogenesis and expand exponentially as the plant grows. By grain development and plant maturity, a root system can have millions of roots. Few studies manipulate more than one stress factor, and few molecular mechanisms have been determined from field-grown roots which continuously encounter multiple soil conditions through space and time. Today the gap between Darwin's carefully controlled experiment and Weaver's field observations remains.
Here we review effects of soil conditions on cereal root architecture. We place particular emphasis on studies on the monocots because cereal species belong to this family and provide >50% of human calories.

Plant root systems
Cereals are members of the grass family, with a monocotyledonous root system comprised of primary axile root(s) from the embryo in the seed, and axile nodal roots from nodes along the stem (Fig. 1). Depending on the cereal, the number of primary roots varies (e.g. from one in the warm climate cereals maize, rice, and sorghum, to six or seven in the cool climate cereals, triticale and wheat). The number and position of emergence of the earliest axile nodal roots also vary, depending on whether the cereal forms a mesocotyl, planting depth, and genotype. Nodal axile roots from the scutellum node from within the seed, and nodal axile roots from the coleoptile node above the seed, may appear as primary roots and collectively be termed 'seminal' roots. Anatomical observations show that primary roots from the embryo have a central, large metaxylem vessel surrounded by small metaxylem vessels, while the basal region of nodal axile roots have a central pith of parenchyma cells surrounded by large and small metaxylem vessels (Watt et al., 2008). Both axile types develop branch roots, which in turn can develop successively higher order branched roots from their pericycle, up to third order. See table 2 in Fitter (1994) showing the different growth rate, length, and longevity of cereal roots of different branching order. Cereal roots, as with all grasses, do not have a vascular cambium, and must develop more root length to generate new xylem and phloem tissues for water and sugar transport, respectively (Fig. 1A). The embryonic and nodal root systems that develop over time offer some flexibility over the lack of cambia because they emerge over time to re-colonize the soil profile and differ in their responses to soil stimuli. The dicotyledonous root system, such as that of Arabidopsis, Medicago, and bean, has a single primary axile root (tap root) that emerges from the base of the embryo. This tap root develops successive orders of branch roots (Fig. 1B). Thus the dicotyledonous root system is simpler than the cereal root systems, with only one type of axile root. 'Adventitious' roots can emerge directly from the stems in dicotyledons, but their contribution to the root system depends on genotype, age, and soil conditions, and they may not be as predictable and substantial as the cereal node axile roots. In contrast to the monocotyledons, dicotyledons have vascular cambia. Vascular cambia divide to differentiate into xylem and phloem tissues, continually thickening the root and providing additional transport capacity.
Despite obvious developmental differences between the monocotyledonous and dicotyledonous root systems, functional differences have not been studied intensively. Hamblin and Tennant (1987) showed in field crops that wheat had longer root length than lupin per unit time, but that water uptake per unit length of lupin was higher than that of wheat. Bramley and colleagues (2009) suggested that this was in part due to anatomical differences, whereby the vascular thickening from the cambium of lupin roots increased water uptake per unit length, while wheat without a cambium needed additional length to add vascular capacity. Nagel et al. (2012) compared the two-dimensional distribution of Arabidopsis roots with the grass model Brachypodium in 60 cm deep root observation boxes. This careful study showed that Brachypodium had greater root length in the middle and deep layers, suggesting that nodal root growth enables denser roots at depth when compared with basal branch roots from the Arabidopsis tap root. Brachypodium provides an opportunity to understand the extent to which processes discovered in Arabidopsis can be applied to cereals because it is a small, easy to grow monocotyledon molecular model (Chochois et al., 2012).
Processes underlying root architecture: tropism, root elongation, and branching Root architecture is a term that describes the pattern of distribution of roots within the soil profile through space and time (Lynch, 1995). Root processes affecting individual roots are what underlie root system architecture, and can be termed 'component traits'. These are (i) branching rates; (i) tropism; and (iii) elongation rates. Together, over time, these component traits determine the architecture of the entire root system. These are also the underlying processes attributed to shoot architecture where they can be explained largely by auxin movements determined in laboratory condition experiments (Prusinkiewicz and Runions, 2012). However, resultant shoot and root architecture in the field are a combination of internal and external signals. Soil conditions are much more complex than aerial conditions experienced by shoots. For simplicity and length, this review deals with abiotic soil conditions and root architecture. Biotic factors are not considered here, although it is recognized that soil organisms such as root-infecting disease, symbiotic organisms, and other rhizosphere organisms have a large effect on root architecture.
Soil conditions and root system architecture: a review of single factors imposed in controlled-environment experiments A root system is predisposed to a shape due to its underlying genetics; however, as soon as roots emerge, soil conditions exert an influence (Fig. 2). A root constantly experiences gravity ( Fig. 2A). Layered over this are soil characteristics ( Fig. 2B), water content ( Fig. 2C), oxygen availability ( Fig. 2D), patches of nutrients (Fig. 2E), and temperature changes (Fig. 2F). Gravity is static over a plant's lifespan, but the other stimuli are heterogeneous both spatially and temporally, and are subject to local depletion or replenishment. In this section, we provide an overview of how root systems respond to these stimuli, resulting in changes to root architecture.

Gravity
Gravity exerts itself constantly and uniformly on all plant organs over the life cycle of the plant, unlike most environmental stimuli. The cell biology of root gravitropism has received a large amount of research attention in controlled conditions. As Darwin and Darwin (1880) hypothesized, the tip is considered the part of the root which senses gravity. The prevailing hypothesis is that dense starch-filled amyloplasts (statoliths) sediment to the lower side of specialized polar cells (statocytes) which are located in the columella of the root cap, and this is sensed by the root (Chen et al., 1999;Boonsirichai et al., 2002). Other mechanisms of gravitational sensing have been proposed: differential pressures within cells are caused by the settling of the protoplast within the extracellular matrix (e.g. in rice, Staves et al., 1997); gravitropic signals move apoplastically within the root cap mucilage from the root cap to the main root (e.g. in maize, Moore and McClelen, 1989); and a signal within the elongation zone contributes to the root gravimetric response (e.g. in maize, Wolverton et al., 2002). Further evidence for these alternative theories is needed, as current evidence generally supports the starch-statolith hypothesis (Boonsirichai et al., 2002;Blancaflor and Masson, 2003).
The signalling cascade that transduces gravimetric information to altered growth once the position of the statoliths is perceived is not completely resolved (Sack, 1991). However, the Cholodny-Went model is widely accepted. It proposes that auxin is redistributed across the elongation zone in gravistimulated roots. Auxin accumulates on the lower side, resulting in differential growth inhibition, leading to downward curvature (Chen et al., 1999;Blancaflor and Masson, 2003;Swarup et al., 2005). Detailed biophysical measurements of cells in the elongation zone of graviresponding roots are lacking in cereals; therefore, it is unknown if the difference in growth rates results from changes in tugor pressure or from asymmetric changes in cell wall properties (Pritchard, 1994). In maize roots, the acceleration of growth on the upper side of the root is accompanied by enhanced acid efflux along the upper surface of the root and a reduction in acid efflux on the lower surface. Therefore, acid-induced wall loosening on the upper surface may account for the downward curvature of the root (Mulkey and Evans, 1981;Versel and Pilet, 1986). Maize has also been used to examine the role of calcium in the root gravitropic response. The upper and lower surfaces of a maize root showed significantly different concentrations of apoplastic calcium, potentially resulting in proton exchange with the stele on the lower side, giving reduced cell wall acidification and, therefore, tighter cell walls and reduced cell elongation on the lower side (Björkman and Cleland, 1991). Calcium is also an essential element in polar auxin transport and has been demonstrated in maize to form gradients in maize root tips when gravistimulated (Lee et al., 1983a, b), potentially resulting in auxin asymmetry (Young et al., 1990) in the elongation zone, resulting in gravitropic curvature. Although little recent work has been conducted on the mechanism of monoct gravitropism, in Arabidopsis, it has been elucidated that root gravitropism is regulated by a transient lateral auxin gradient (Band et al., 2012b) The angle at which roots emerge ( Fig. 2A) is known as the 'gravitropic set-point angle' (Digby and Firn, 1995). It has a genetic component (Uga et al., 2011;Mace et al., 2012) and has been used as an early screen for root traits in cereal breeding programmes (Wasson et al., 2012). The root angle at which primary roots emerge has been linked to the depth of the mature roots system in the field for several cereal species (e.g. rice, Kato et al., 2006;sorghum, Mace et al., 2012;wheat, Oyanagi et al., 1993b;Manschadi et al., 2008). The association between deeper rooting and root angle has been exploited in rice, leading to the discovery of quantitative trait loci (QTLs) which appear to correlate strongly with a narrower and potentially deeper rooting phenotype (Uga et al., 2011). Nodal root angle QTLs have also been identified in maize (Omori and Mano, 2007) and sorghum (Mace et al., 2012), where three QTLs show homology to the previously identified root angle QTLs in rice and maize.
Although gravity is distributed uniformly across roots in soil, and hence gravitropism is relatively straightforward to target for crop improvement, the actual directions of growth of roots in soil are variable. They depend on root type, soil condition, and genotype. For example, primary, leaf node, and branch roots of cereals behave plagiogravitropically. They do not grow directly downwards, but, after emergence, can elongate horizontally for some time before curving vertically downwards (Nakamoto et al., 1991;Oyanagi et al., 1993a). Root angle can be influenced by temperature and water availability (Oyanagi et al., 1993a). Oyanagi et al. (1992) found that the angle from the horizontal of wheat primary roots increased with water stress in 50 of 133 genotypes tested. The remaining 63 lines did not show significant change from the angle of well-watered plants, indicating a genetic by environment interaction. The use of the root angle in response to gravity as an early screening proxy for deeper root systems may need to include additional screens for responses in nodal and branch roots, and responses in the presence of water and other soil conditions, for improved correlation to the field. The mechanisms by which root tips perceive multiple stimuli such as gravity, water, and temperature, and integrate those for a resultant curvature, are not known, but could be critical for field crop improvement.

Soil structure
Soils are triphasic, consisting of solid particles, liquid water, and air. The solid phase constitutes ~50% of the total soil volume, and the texture of the soil is determined by the size of these particles. Soils will often contain numerous different particle sizes, with this composition having a significant effect on the soil structure (i.e. the size of particle aggregates and the resultant soil porosity). Both soil texture and structure impact on bulk density and on how easily a soil will be compacted, in turn contributing to soil strength. Moisture content also influences soil strength, which increases with decreasing soil water, due to the matric potential becoming more negative as a result of capillary forces (Whalley et al., 2005). These soil conditions are highly influential for root architecture, impacting the mechanical impedance (physical stress) to root elongation through the soil, as well as affecting the availabilities of water, oxygen, and nutrients (Gliński and Lipiec, 1990;Gregory, 2006) (Fig. 2B).
Roots encounter some amount of mechanical impedance to elongation as they grow through the soil. The main influence of higher impedance is a slowing in the rate of root elongation, with a coinciding increase in root diameter (Bengough and Mullins, 1990;Gliński and Lipiec, 1990;Gregory, 2006). Root elongation decreases in an approximately linear fashion until very high resistance is encountered (Whitmore and Whalley, 2009). For example, Merotto and Mundstock (1999) grew wheat at 1.0, 2.0, 3.5, and 5.5 MPa and found that as soil resistance increased, from as early as 16 d after emergence there was a significant difference in root growth between the four soil treatments, with roots in stronger soils showing a heavily reduced length, surface area, and dry matter, but having a higher root diameter. This has also been shown in numerous other cereals (e.g. oats, Ehlers et al., 1983;maize, Goodman and Ennos, 1999). The effect of soil mechanical impedance on lateral root growth in cereals has received less attention, and there is not agreement about how lateral roots are affected. Bingham and Bengough (2003) found that variation in soil strength had no effect on the density of lateral roots (number per unit main axis length) in either barley or wheat, although in stronger soils elongation was repressed in lateral roots and similar results have been found in oats (Schuurman, 1965). Primary wheat roots in dense soil had decreased elongation but accelerated initiation of branch roots compared with those in loose soil (Watt et al., 2003). Other authors, however, uphold that branching density is also affected by soil strength (Goss, 1977;Iijima and Kono, 1991). Variation in responses in branch roots may be due to interactions with aspects of soil texture such as soil pore size (discussed further below) and clod density (Konôpka et al., 2008), which can strongly affect lateral branching density. Shorter total root length in strong soil results in less effective soil exploration, resulting in lower water and nutrient interception, consequentially diminishing supply to the shoot, which can result in reduced total plant growth on compacted soils.
Soil strength and mechanical impedance are rarely uniform. Soils can have subsoil hardpans, cracks, and macropores due to drying and areas of higher compaction. Hardpans can decrease rooting depth, although this does not necessarily correlate with a lower overall root mass as plants may respond with increased root exploration above the hardpan (Barraclough and Weir, 1988). In areas of uneven compaction, such as under tractor wheel tracks, roots will tend to be significantly less dense where compaction is highest (Tardieu, 1988). The previous two examples demonstrate that roots systems will favour areas of least mechanical resistance, and this is especially highlighted in soils containing cracks or pores (Fig. 2B). Roots will favour the low impediment pathway formed by root channels from previous crops or cracks. For example, in a maize crop planted in rotation after alfalfa, 41% of the maize roots were found in alfalfa root channels (Rasse and Smucker, 1998). White and Kirkegaard (2010) found >90% of wheat roots deeper than 40 cm clustered in cracks and pores in red brown clay field soil. The exploration of soil cracks and pores may allow roots to penetrate hardpans, or to explore deeper areas or soil, thereby gaining access to larger pools of water and nutrients (Passioura, 1991).
Root extension occurs when the tugor pressure within cells of the elongation zone becomes sufficiently large so as to overcome the constraints imposed by cell wall resistance and resistance of the surrounding medium (Pritchard, 1994). Roots easily elongate through the air and liquid phases of the soil. Where the solid phase dominates, roots must exert enough pressure to displace soil particles and overcome friction, allowing elongation (Clark et al., 2003). It has been demonstrated in maize that mechanical impedance results in increased ethylene biosynthesis (Moss et al., 1988;Sarquis et al., 1991); however, the action of ethylene in root sensing and response to soil impedance is not yet understood.
The genetic basis of the ability of roots to penetrate soil is not completely elucidated for any species, and in the cereals has received limited attention. Studies of rice and maize have shown strong genotypic differences in root penetration capacity (Yu et al., 1995;Bushamuka and Zobel, 1998), Soil conditions and root architecture | 1197 indicating a genetic basis to this ability. In rice, several QTLs have been identified using a wax penetration screening system (Ray et al., 1996;Zheng et al., 2000). Root penetration is a complex trait and it is possible that differences in root thickness, allowing roots of different genotypes to resist buckling, could be a major contributor to better root penetration; this is supported by evidence that QTLs for the ability of roots to penetrate a wax layer are associated with QTLs which make the roots either thicker or longer in other studies (Price et al., 2000).
Given the propensity for roots to grow in cracks and spaces rather than push through hard soil, it may be beneficial to select for inherent rapid elongation rates of root systems, to increase colonization of dense soils in the field (Watt et al., 2005;Palta and Watt, 2009). This approach should, however, take into careful consideration the size of the pores into which the roots will grow. Plants with roots in large pores can have poorer shoot growth compared with plants with roots in compacted soil (Stirzaker et al., 1996; shoots of roots shown in Fig. 2B; Garcia, Watt, and Passioura, unpublished). The mechanisms by which shoot growth is inhibited by roots growing in large pores are unknown, but may include inhibitory signals from the root tips that accumulate because of lack of soil contact, or inhibitory biotic factors in pores (Passioura and Stirzaker, 1993).

Soil water content
Water deficit is the most prevalent limitation to plant growth. Severe water stress, when the water potential is significantly lower within the soil than in the plant, will result in dehydration and loss of cell tugor, usually resulting in severe restriction of root elongation (Fig. 2C, inset). However, the often heterogeneous nature of water within the soil profile and the plastic nature of root systems means that while prolonged periods of water stress are generally detrimental to overall plant health and crop yields, many species can withstand extended periods without water inputs.
Under limited water conditions, tolerant plants tend to grow a deeper root system, prioritizing assimilates from shoot growth to root growth so roots can extend into still moist deeper zones (Gregory, 2006). A deeper root system is achieved by the ability in many species for root systems to continue to grow, even at extremely low water availabilities (Sharp et al., 2004). Maintenance of root elongation when water stressed allows for continued soil exploration and, therefore, increased chances of maintaining adequate water supply to the shoots (Ingram and Malamy, 2010). In primary roots, this feature could be vital in seedling establishment when topsoils are dry (Sharp et al., 1988), and in nodal roots this may be important for penetration of dry surface soils. Westgate and Boyer (1985) demonstrated this ability in the nodal roots of maize which continued to elongate several days after water was withheld; root elongation was still occurring at water potentials of -1.4 MPa, whereas leaf elongation ceased at -1.00 MPa. Although root elongation can be maintained during periods of water stress, and new nodal root production may continue (Pardales and Kono, 1990), roots developed while under water stress will usually be thinner (Sharp et al., 1988) and the rates of elongation and branching are usually lower than those in wellwatered soils (Salim et al., 1965;Stasovski and Peterson, 1991). However, in some cases, elongation rates in waterstressed roots have exceeded that of plants with sufficient water (e.g. maize, Sharp and Davies, 1979). Although roots can maintain growth at matric potentials which inhibit shoot growth, under extremely low water potentials (i.e. less than -1.5 MPa) root growth will be inhibited. This is demonstrated in a pot experiment where water availability was stratified down the pot, and for wheat, oats, and barley most root growth stopped as soon as roots reached the dry (2.3% moisture content) soil layer (Salim et al., 1965). Continued elongation under water stress probably relates to the hydrotropic nature of roots which may allow them to bend towards areas of higher water potential (Takahashi, 1997;Eapen et al., 2005). Although it is likely that all roots show hydrotropism, it has received little attention in the cereals; however, Takahashi and Scott (1991) showed root hydrotropism in maize and concluded that the hydrosensory site was located within the root cap.
Our understanding of the physiological mechanisms involved in the maintenance of root elongation under water stress is based primarily on studies of short-term water stress on the primary root of maize seedlings (reviewed by Sharp et al., 2004). These experiments reveal that the root growth zone behind the tip is shortened by water stress, from 12 mm in a well-watered root to 7 mm under water stress, but the time over which growth occurs is independent of water stress (Sharp et al., 1988). Maize roots undergo substantial osmotic adjustment during water stress (Sharp and Davies, 1979;Sharp et al., 1990); however, this is insufficient to maintain enough tugor for growth. Therefore, it appears that cell wall-loosening proteins such as expansins are involved in the maintenance of elongation in the root apices of waterstressed roots (Wu et al., 1996). Hormonal involvement in root elongation during water deficit is not well understood. The roles of abscisic acid (ABA) and ethylene have received some attention. although their roles are yet to be fully elucidated. ABA has been shown to accumulate in the root growth zone at low matric potentials, leading to restricted ethylene production, and this interaction may function to maintain root growth (Sharp, 2002).
When one part of the root system receives water, other parts in dry environments can elongate (Boyer et al., 2010). Working with wheat primary roots, these authors showed that roots extend into extreme dry conditions (in air) provided another part of the root system has water. These experiments were used to suggest that ~50% of the water used for wheat primary axis extension came from the phloem; and the remaining 50% came from the soil. In maize, the proportion of water for root growth from the phloem was estimated to be 80% (Brett-Harte and Silk, 1994). Anatomical or transporter processes associated with the phloem could be manipulated to increase the efficiency of water phloem transport to roots to maintain root extension as the surrounding soil dries.
Drought tolerance is a primary aim in breeding, and this has resulted in the genetics of cereal root growth in response to water stress receiving attention. In rice, for example, numerous QTLs have been identified under water stress for root traits such as seminal root length, root diameter, total root length, or mass (Champoux et al., 1995;Yadav et al., 1997;Zhang et al., 2001;Price et al., 2002).

Anoxia and hypoxia
Excess soil water is a common problem at transient or seasonal time scales (Kozlowski, 1984). Waterlogging results in soil hypoxia (low oxygen) or anoxia (no oxygen) of the rhizosphere within hours (Ponnamperuma, 1984). Even shortterm (hours or days) transient root exposure to hypoxic soils can have considerable negative effects on growth and yield of dryland crops (e.g. wheat, Davies and Hillman, 1988;barley, Leyshon and Sheard, 1974;Sharma and Swarup, 1988;millet, Sharma and Swarup, 1989). Although the responses of root systems to excess water are highly variable, a combination of reduced primary root growth in anoxic sediments and root decay generally reduces the root to shoot ratio (R:S). Even flood-tolerant paddy rice has a low R:S of 0.13-0.23 depending on variety and age (Teo et al., 1995), compared with 0.4-0.55 in dry-land cereals such as wheat when in drained soil (Siddique et al., 1990).
Excess anoxic water around roots is generally detrimental to primary cereal roots, inhibiting elongation and lateral root initiation. When wheat seedlings were returned to aerated conditions after 10-14 d waterlogging, seminal roots of wheat did not recover or resume elongating, indicating either death or serious injury (Barrett-Lennard et al., 1988;Malik et al., 2002). With shorter periods of waterlogging, seminal apices appear to have died but lateral emergence from the basal zones of seminal roots did occur once roots were returned to aerated conditions (Barrett-Lennard et al., 1988). Similar root responses have been observed in waterlogged maize roots. After waterloggimg for 12-13 d, seminal root elongation either ceased or slowed significantly and there was also significant root senescence (Wenkert et al., 1981;Przywara and Stepniewski, 1999). New nodal root growth, however, can also be promoted by waterlogging (Fig. 2D) (Trought and Drew, 1980;Wenkert et al., 1981;Erdmann and Wiedenroth, 1986;Lizaso et al., 2001;Malik et al., 2002;Abiko et al., 2012). These new nodal roots can be thicker (due to aerenchyma formation) and shorter than pre-existing roots (Malik et al., 2001;Abiko et al., 2012), and branching patterns can also be different. The survival of these nodal roots after waterlogging varies between species but, if maintained, can influence future root system architecture.
In dicotyledons, auxin and ethylene are accepted to be the hormones associated with initiation of new adventitious roots after soil waterlogging (Visser and Voesenek, 2004). They are thought to interact; ethylene modulating auxin transport and biosynthesis, and sensitizing plant perception of auxin (Visser et al., 1996). However, for monocotyledons, the hormones involved in adventitious root initiation are less clear, as data are limited and somewhat contradictory. Zhou et al. (2003) found polar auxin transport essential for adventitious root initiation and elongation in rice, while Lorbiecke and Sauter (1999) found no effect of applied auxin on adventitious root formation in rice. Little experimental evidence exists for the role of ethylene in cereals; however, in rice, it has been demonstrated to induce adventitious root growth (Lorbiecke and Sauter, 1999) where it may be involved in facilitating penetration of the epidermis and thus emergence of adventitious roots from stem nodes. High concentrations of ethylene resulted in programmed cell death of epidermal cells of the node, just external to the tip of adventitious root primordium (Mergemann and Sauter, 2000). Several studies have looked at a possible role for gibberellins in adventitious root formation in rice, but results are contradictory (Suge, 1985;Lorbiecke and Sauter, 1999).
The genetic basis of adventitious root formation as a response to excess water is still not completely understood, and in the cereals little has been elucidated. In teosinte (Zea mays ssp. huetenangensis) a QTL on chromosome 8 has a strong association with adventitious root formation (Mano et al., 2005), and in rice cdc2Os-1 has been shown to be expressed exclusively in developing adventitious roots, and may, therefore, be part of a root-specific signal transduction pathway triggered by ethylene (Lorbiecke and Sauter, 1999).

Nutrients
Root architecture determines access to nutrients in soil, and the underlying processes-curving, elongating, and branching-can all be influenced by internal and external nutrient availability. For example, limited plant-available phosphorus is associated with a more horizontal root angle in bean, placing roots in surface soil where phosphorus can accumulate because it is highly immobile (Bonser et al., 1996). Root elongating and branching can be triggered by local nutrient abundance (Robinson, 1994). The most well known of the numerous studies examining root responses to nutrient heterogeneity are those by Drew (1975). Via destructive harvests, Drew (1975) demonstrated that in barley, grown for 21 d in pots stratified with different concentrations of phosphate, nitrate, or ammonium, seminal roots initiated many new firstand second-order laterals in zones supplied with high nutrient concentration. Micro-computed tomography now allows nondestructive illustration of lateral root proliferation in nutrient patches (Fig. 2E). Root proliferation within patches of high nutrient availability can result from changes to branching or elongation. The angle at which lateral branches emerge will affect the area over which the roots spread, so narrower branching angles will create a denser patch, especially if the root also changes its branching pattern (Fitter, 1994). Higher root proliferation can also be reflected in high nutrient soils; however, for some nutrients such as nitrate, high concentrations can inhibit root elongation in some genotypes (e.g. maize, Tian et al., 2005;Wang et al., 2005). 'Plasticity', that is, the extent of root architecture response to local or general nutrient changes in the soil or plant, due to plant genetics, may be an advantage where nutrient distribution is heterogeneous through the season and with depth.
Extreme nutrient deficiency will hinder root system growth, but under less severe nutrient deficiency carbon is allocated away from shoots and towards roots, leading to root elongation (e.g. rice, Kirk and Du, 1997;barley, Steingrobe et al., 2001). Nutrient uptake, particularly for immobile nutrients such as phosphorus, is restricted by the root surface area available for uptake, and elongation increases soil exploration (Manske and Vlek, 2002). In a pot experiment with nine genotypes of wheat, low phosphorus supply significantly increased the total root length by 56% (Römer et al., 1988). Maize had a similar response when exposed to nitrogen deficiency, increasing total root length and lateral root growth (Chun et al., 2005).
Knowledge of the signalling and genetics of root architecture responses to nutrient availability is limited largely to Arabidopsis, with a focus on lateral root development and inhibition (Osmont et al., 2007;Nibau et al., 2008;Ingram and Malamy, 2010). In the cereals, QTLs for root responses to phosphorus or nitrate levels have been identified in maize (Zhu et al., 2005;Liu et al., 2008) and rice (Shimizu et al., 2004).

Temperature
Soil temperature is highly dynamic, and can differ from the temperature experienced by shoots. It fluctuates daily, seasonally, and with depth. Deeper soils are insulated and fluctuate much less than the surface, and the temperature range around the root system depends on location (Fullner et al., 2012).
Soil temperature directly affects root elongation, metabolism, and transporters, and indirectly affects shoot processes. Studies in controlled environments show that roots generally elongate more rapidly as temperature warms (Fig. 2F), reaching a peak after which rates decrease (Gregory, 2006). Abbas Al-Ani and Hay (1983) reported that barley, rye, oat, and wheat seminal root elongation increases linearly from 5 °C to 25 °C, and suggested that the fastest rates for these cereals lie at or above 25 °C. However peak rates may depend on other experimental conditions. For example, wheat grew most rapidly at 16 ± 3.7 °C (Porter and Gawith, 1999). Among cereals, the temperature at which roots grow the quickest is wide; maize radicals were quickest at 30 °C (Blacklow, 1972), and oat radicals at 5 °C (Nielsen et al., 1960). However, temperatures in older studies may not have been well controlled and peak rates may depend on root age (Gliński and Lipiec, 1990). As well as elongation, branching can also be affected. Nagel et al. (2009) showed in maize that at 14 °C (as opposed to 24 °C) root biomass was reduced, particularly in the laterals. Interestingly Burström (1956) found in wheat that while individual root elongation rates increased with temperature, the overall root length actually decreased. Although root elongation rates increase with increasing temperatures, an increased development rate within individual roots can result in smaller cells, due to a decreased cell elongation phase. Further, correlations between temperature and root diameter have been shown to be both positive and negative (Nielsen et al., 1960;Pahlavanian andSilk, 1988), or unaffected (Abbas Al-Ani andHay, 1983). Inconsistency in thickness responses to temperature may be due to methodology, age measurements, branch order, or types of roots used in different studies.
Changes in root growth are intrinsically linked to temperature mediation of other root processes (Gliński and Lipiec, 1990). Water and nutrient uptake by roots increases with temperature (up to a point). Warm soils may correspond to warmer shoots and, therefore, higher photosynthetic rates, providing more substrate for growth (Takeshima, 1964;Chapin, 1974;Frossard, 1985;Borisova, 1991); alternatively, soils may be cooler than the shoot atmosphere, resulting in negative feedback to the shoots. Root to shoot communication regulating soil temperature effects on roots and shoots is poorly understood, yet has wide implications for crop productivity.
Soil conditions on root system architecture: review of multiple factors imposed in controlled-environment experiments In contrast to a root in laboratory experiments where most often one stimulus is manipulated (Fig. 2 examples), a root in the field is exposed to multiple stimuli at one point in time (Fig 3). Over the life of a field root system, stimuli such as Fig. 3. Schematic of the interacting soil conditions which also impact on processes within a root. These soil conditions will influence the individual root length and development, and therefore the overall root system architecture.
gravity are static, but most others such as temperature, nutrients, and water fluctuate diurnally and seasonally around each root as it elongates, and its rhizosphere is depleted and replenished over time and space. How do roots sense and respond to multiple soil conditions of the field? Are resultant root architecture and function a net, sum, or compound response? While there are some controlled-environment studies testing interacting effects of two soil conditions on roots, studies beyond a single interaction are scarce, even of dicotelydons.
Studies of interactions between temperature and other soil conditions are relatively common, possibly because it is relatively straightforward to impose a uniform temperature difference in an experiment. These studies tend to show warmer soil reducing growth restrictions by other stimuli. For example, Abbas Al-Ani and Hay (1983) demonstrated that higher temperatures compensated for detrimental effects of mechanical impedance in four cereals growing in four media of different densities and temperatures. However, when waterlogging was added to impedance experienced by maize roots, warmer temperatures did not increase penetration depth; rather they increased waterlogging-induced senescence compared with roots in cooler soils (Przywara and Stepniewski, 1999). In drier conditions, a 15 °C temperature increase doubled water absorption by rice roots (Takeshima, 1964). Higher temperatures also increased water uptake in maize roots (Frossard, 1985;Borisova, 1991). Nutrient uptake also tends to be higher at warmer temperatures (e.g. barley, rice, Takeshima, 1964). Hussain and Maqsood (2011) recently found that a 4.2 °C increase around maize roots increased uptake of P, Cu, and Zn per root dry matter. Temperature has also been implicated in the gravimetric response of roots in several species. For example, maize nodal root angle was most shallow at 17 ºC, becoming more vertical in warmer or cooler soil. Interestingly, Onderdonk and Ketcheson (1973) imposed fluctuating temperatures, as would be found in the field, and these resulted in a similar root angle to the warmest point in the temperature cycle.
A few studies have examined the interaction between moisture stress and other soil conditions. For example, nodal roots of maize grow more vertically in drier conditions (Nakamoto, 1993). In an elegant study using a miniature tension table with Arabidopsis lines reported to have differences in nitrogen responses, Chapman et al. (2011) found that small changes in water-filled porosity of sand (e.g. water around the roots) stimulated tap root growth, and removed induction of branch root growth by nitrate previously characterized on agar.
Some soil conditions are intrinsically related, and root responses cannot be attributed to a single stimulus in controlled or field experimental conditions. Water and oxygen are an example. Both are essential for healthy root growth. In soil, they are negatively correlated; as soil dries, water in pores decreases and oxygen diffusion and availability increase. However, lower water increases soil strength due to capillary forces, increasing mechanical impedance to roots (reviewed in Bengough et al., 2011). For optimal root growth, a balance must be achieved between water availability, oxygen diffusion, and mechanical impedance. The effect of water content on aeration and impedance is strongly affected by bulk density and soil texture.
In controlled-environment studies, interactions between the growth media and the characteristics of interest need to be considered. This is especially relevant for studies using the model Arabidopsis, often on agar. In investigating the interaction between water supply and nitrate on Arabidopsis root growth, Chapman et al. (2011) found that, with the exception of lateral root number, root architectural characteristics varied significantly for seedlings grown on sand as opposed to agar. Root growth responses to changes in nitrate supply were also different depending on growth media. As well as differences in mechanical impedance and root gas exchange, the hydraulic properties of sand and agar are very different, as are those of different concentrations of agar (a 0.1% change in agar concentration can significantly alter the water matric potential; Spomer and Smith, 1996). Given that this determines delivery of water and nutrients to the root, it is likely that properties of the agar would create a significant interaction in water or nutrient use studies (Chapman et al., 2011) Root architecture in the field Although the controlled environment has limitations for root studies, the field is not without its restrictions due to fieldsown trials experiencing all soil parameters simultaneously and all possible interactions between these parameters. While some manipulation of the soil condition is possible (fertilizers, rainout shelters, soil preparation), much of the soil environment is beyond the researcher's control. For example, Fig. 4 shows two field sites, situated on the same farm, <1 km apart, which were sown with various cereals in 2011. Site preparation and treatment were the same; however, plant roots respond to the sites very differently. The sites were cored to 2 m at maturity and core break counts were used to assess root depth and distribution. Rapid measurement of root architecture in relation to depth in the field is most commonly undertaken using soil cores of 4-10 cm diameter taken below the crop with the core break method (e.g. Fig. 4A, B). In this method, the number of roots observed on the broken faces of a core are counted and then a selection of samples are washed out to calculate a calibration of the method from the correlation of root length density or root dry mass to core face root counts at different depths. At best, destructive root counts are then correlated to multiple soil conditions taken at the time of root coring (Fig. 4C-F); however, root and soil cores are often taken at different times: soil analyses at the beginning of crop growth; roots at the end.
In the example in Fig. 4, significant genotype×soil condition variation is observed. Wheat cv. Westonia root systems are deeper and more dense at depth than those of cv. Beaufort at soil Site 1 (Fig. 4A, B), but of a similar depth but less dense at soil Site 2. Soil analyses confirm that Site 1 is less dense (Fig. 4C), and with higher water (Fig. 4D) and nitrate (Fig. 4B) holding capacity than Site 2, possibly suggesting that the Westonia is more sensitive to the impedance Soil conditions and root architecture | 1201 of Site 2. Westonia is a vigorous variety, and may have exhausted water through excessive transpiration, increasing soil strength around the roots in Site 2. However, this is purely speculation as both soil and root measures are at a single and separate point in time. The soil analyses, however, do not explain Beaufort's relatively poor root growth in Site 1. This illustrates the challenge faced by root scientists working in the field; it is currently impossible to quantify all factors that could potentially be affecting root growth, and the interactions between them are even more difficult to characterize. Although these root and soil analyses are crude, they are the first step towards understanding root architecture responses to soil conditions. They are rare in the literature.
The above examples highlight that while single-factor controlled-environment studies are not without merit, the results of such studies should be treated with caution as their translation to field conditions may yield a very different root architectural response due to the broader spectrum of interacting parameters roots will encounter there. The correlation between controlled single-factor studies and the field has beeen investigated by some researchers (Nass and Zuber, 1971;Arihara and Crosbie, 1982;Landi et al., 2001); however, correlations tend to be highly variable and determined by the exact experiments performed and field conditions studied.

Some ideas for linking Darwin and Weaver
From the time of Darwin, our understanding of root architecture in the field has relied on the results of reductionist studies in the laboratory, as illustrated in Fig. 5. However, laboratory conditions are much simplified compared with pots and soil conditions in the field. Compounding this are differences in the roots themselves, with older plants containing Fig. 4. Root counts of two wheat varieties (A, B) and soil bulk density (C), moisture (D), nitrogen content and pH at two sites. Site 1 (filled circles) was situated on the same farm, <1 km from Site 2 (open circles) and both were sown with various cereals in 2011. The sites were cored to 2 m at maturity, core break counts were used to assess root depth and distribution, and the soil was sampled at maturity to determine soil characteristics. Site 1 n=4, Site 2 n=5. different root types and ages, and varying in density with depth ( Fig. 5A-C).
Three research disciplines allow two-way scaling in root research, between controlled and field environments. First, models could be parameterized with ranked effects of soil factors on root architecture. Letey (1985) suggests the concept of the 'non-limiting water range' as a way of discussing the relationship between soil water content and other soil factors which constrain plant growth. Letey (1985) defines the 'nonlimiting water range' as being within the range of water availabilities at which the plant will grow and then the lower end of this range is increased to a point where mechanical resistance is not limiting and the upper end of the range reduced to a point where aeration is not limiting. Depending on the soil type, this range can be similar to that given by plant-available soil water, or could be far smaller. As well as interacting with oxygen availability and soil strength, water availability will also interact with the uptake of most nutrients. For example, in wheat, low water availability decreases the R:S, whereas low N availability results in an increased R:S; the interaction of low water availability and low N results in the increased root exploration response to low N being repressed and the root mass being lower than in controls (Barraclough et al., 1989). In Table 1, we hypothesize the importance of various environmental stimuli on root architecture processes. While our assessment is primarily drawn from controlled-environment and glasshouse studies, we believe rankings such as these could be used in combining traditional soil measurements with controlled-environment knowledge of effects on root architecture cellular processes (curving, elongation, and branching) to estimate root architecture for a given soil and management practice. Mechanistic models of root growth have primarily described the regulatory processes involved in root development, often at a single root scale, and usually on dicot model species (Dupuy et al., 2010). These models therefore, while providing a good foundation, are of somewhat limited usefulness due to the significance of environmental stimuli to root growth and architecture. Recent models are making forays into incorporating environmental factors (Roose and Schnepf, 2008;Draye et al., 2010). The inclusion of rhizosphere and environmental data into mechanistic models and the production of environmental data able to be incorporated into multiscale models (Band et al., 2012a) will result in both a greater understanding of the genotype to phenotype gap and as a guiding framework for new experimental work.
Secondly, our understanding of the biology of sensory and signalling pathways at the root-soil interface could be greatly expanded. Plants with reporter systems can be studied in non-invasive measurement systems for roots in intact soil methods, such as X-ray tomography (e.g. Flavel et al., 2012;Fig. 2E) and magnetic resonance imaging (MRI) (e.g. Jahnke et al., 2009). Real-time, high-resolution root-soil methods are the beginning of understanding root architecture in the field; however, as yet, most of these technologies are restricted to pot-based studies. Small model plants growing to maturity in intact soil [such as the model grass Brachypodium where adult stages can be studied and selected in soil columns to 50 cm (Chochois et al., 2012)] can increase our understanding not only of root architectural phenotypes but also of the sensory and signalling pathways which determine root growth response to soil conditions. These systems have also yet to be exploited to understand sensory and signalling processes, which are likely to be the key to genetic improvement by gene marker-based breeding. Crop plants with disruptions known to be in signalling pathways (tilling, mutagenized, sequenced) can also be studied in these non-invasive systems that allow 4D root measurements.
The third, and most important, discipline to link Darwin and Weaver is agronomy: the science of soil management and crop production (Oxford English Dictionary). This sounds like an old-fashioned science, but it is the only to cover the linking of soil and root architecture for crop production gains. Soil conditions and root architecture | 1203 Underlying agronomy today are technology developments in plant genetics, computing power for models, and engineering technologies. Better real-time imaging technologies will enable better understanding of root responses to multiple soil conditions and genotypic root architectural phenotypes, and allow researchers to work readily from the field to the lab and back. Table 1. Overview of the hypothesized importance of various environmental stimuli on root architecture processes. Evidence in support of our hypothesized combinations of traits is given in the text.