Abstract
Root hairs are important sites for nutrient uptake, especially in P limiting conditions. Here we provide first insights into root hair development for the diverse root types of rice grown under different conditions, and show the first in situ images of rice root hairs in intact soil. Roots of plants grown in upland fields produced short root hairs that showed little responsiveness to P deficiency, and had a higher root hair density in the high P condition. These results were reproducible in rhizoboxes under greenhouse conditions. Synchrotron-based in situ analysis of root hairs in intact soil further confirmed this pattern of root hair formation. In contrast, plants grown in nutrient solution produced more and longer root hairs in low P conditions, but these were unequally distributed among the different root types. While nutrient solution-grown main roots had longer hairs compared to upland field-grown main roots, second order lateral roots did not form any root hairs in nutrient solution-grown plants. Furthermore, root hair formation for plants grown in flooded lowland fields revealed few similarities with those grown in nutrient solution, thus defining nutrient solution as a possible measure of maximal, but not natural root hair development. By combining root hair length and density as a measure for root hair impact on the whole soil-grown root system we show that lateral roots provided the majority of root hair surface.
Introduction
In general, nutrient deficiency leads to decreased plant growth, delayed development, and yield loss. Being highly sorbed by soil minerals, the essential nutrient phosphorus (P) is considered immobile (Schachtman et al., 1998), and 50% of crop-planting soils worldwide have been estimated to be P deficient (Lynch, 2011). To overcome yield limitations fertilizer is commonly added, mostly as superphosphate, yet only 10–20% of applied P is efficiently used by plants in the year of application (Holford, 1997).
Identifying and improving traits for more efficient P uptake is an important agricultural goal. Two major strategies to increase P uptake by plants facing P deficiency are to improve P mobilization and to improve soil exploration (reviewed in Lynch, 2011; Rose et al., 2013). The first strategy includes higher P solubilization via root exudation of low molecular-weight organic acids, protons, or phosphatases as well as the increased production of P transporters. The latter strategy leads to an increased root surface area (RSA) by formation of finer roots, aerenchyma, and root hairs (Lynch, 2007). Of these strategies, root hair formation represents the response to P limitations with minimal carbon cost (Lynch and Ho, 2005).
Root hairs, tubular shaped unicellular extensions of epidermis cells, are important for water and nutrient uptake (Gilroy and Jones, 2000), and increase barley (Hordeum vulgare) root anchorage in soil (Haling et al., 2013). Interestingly, while P deficiency leads to formation of longer root hairs in Arabidopsis thaliana, deficiencies in other nutrients have been observed either to have no effect or to decrease root hair length under laboratory conditions (Bates and Lynch, 1996). Direct involvement in P uptake from the soil has been demonstrated for rye (Secale cereale) root hairs (Gahoonia and Nielsen, 1998), and indirectly for wheat (Triticum aestivum) and barley root hairs (Gahoonia et al., 1997). Under low P conditions root hair length and density increase, and Arabidopsis plants with longer root hairs were demonstrated to take up more P (Bates and Lynch, 2000). Modelling approaches identified root hair density to be the less important trait for soil P uptake compared to root hair length (Zygalakis et al., 2011), and to both hair length and longevity (Brown et al., 2012). A study in barley concluded that it was the presence of root hairs rather than their absolute length that was the important factor maintaining yield in P limiting conditions (Brown et al., 2012).
In rice (Oryza sativa L.) P deficiency is an important constraint for growth, and its increased capture from soil presents a cost-effective route to higher efficiency and thus higher yield (Ismail et al., 2007). In comparison to rain-fed lowlands, upland field conditions are more severely affected by P deficiency due to its reactions with iron and aluminium in acidic soils and calcium in neutral and alkaline soils, leading to the formation of poorly soluble complexes (Holford, 1997).
Compared to other crop plants including barley, wheat, and maize (Zea mays) where root hairs are often subject to experiments in controlled solution or soil-based conditions, and sometimes under field conditions, rice root hairs have rarely been investigated. Where they have been investigated, artificial conditions based on nutrient solution (Yu et al., 2015) or MS-agar (Kim et al., 2007) are usually employed, with growth under lowland field conditions being the only agriculturally relevant condition examined (Kawata and Ishihara, 1961). Responses to specific nutrient levels are often studied in artificial, controllable systems such as agar plates and nutrient solutions, raising the question of how representative these measured traits are of roots grown under soil conditions. Also, as root hair measurements are time consuming, only one root type – usually the primary, seminal, or crown root – is generally considered per study, with lateral roots being neglected. The rice root system contains the first emerging seminal roots, shoot-borne crown roots, and two distinct classes of lateral roots (LRs) differing in their morphological and histological structure; long and thick L-type LRs capable of producing higher order LRs, and non-branching, short and thin S-type LRs (Kono et al., 1972; Yamauchi et al., 1987). The approaches published so far usually focus on seminal or crown roots and therefore represent snapshots of root hair formation. In order to better understand root hair formation on a root system scale, we conducted experiments in soil and artificial growth conditions to address the following three questions: (i) do root hair length and density differ depending on P availability in the growth medium; (ii) to what extend does root hair formation vary between main roots and the lateral root types; and (iii) can screening in artificial growth conditions reliably predict root hair properties under field conditions?
Materials and methods
Rice varieties used for the experiments
To investigate general developmental differences in rice (Oryza sativa) roots in response to different media and P levels, a diverse set of varieties was chosen. Example varieties of indica and aus populations were used that are adapted to either lowland (IR64, Taichung Native, Mudgo, Sadri Tor Misri, Yodanya) or upland (Santhi Sufaid, DJ123, Nerica4, Lemont, Kalubala Vee) conditions. Dormancy of seeds was broken by incubation at 50 °C for 2 d.
Nutrient solution growth conditions
Seeds were surface-sterilized in 0.5% NaOCl 0.15M KH2PO4 for 5min followed by three washing steps in distilled water. Germination was initiated in distilled water in petri dishes at 30 °C under dark conditions for 2 d. Seedlings were transferred to mesh trays floating in water in the greenhouse for further growth. Five days after sowing (DAS) iron (12 µM FeEDTA) and calcium (0.1mM CaCl2) were added and the pH adjusted to pH 5.7. At 14 DAS seedlings were transferred to 45 l containers containing third-strength Yoshida solution (Yoshida et al., 1972) at P concentrations of either 1 µM (low P) or 100 µM (high P). Target P concentrations were maintained via the addition of NaH2PO4 2H2O every second day. Individual seedlings were arranged in 3cm diameter holes drilled into container lids and fixed using sponge material wrapped around the stem. After 7 d the solution was replaced by half-strength Yoshida solution followed by full-strength Yoshida solution at 28 DAS. The pH was adjusted regularly to 5.7. Five replicates (per genotype) of all ten genotypes mentioned above were grown in a completely randomized design per treatment. All plants were harvested 35 DAS for root hair investigation. Roots were stored at 4 °C in 50% ethanol prior to microscopic evaluation.
Rhizobox growth conditions
A P-deficient Andosol (1.04 µg P cm−3 plant available P; Wissuwa, 2003) from an upland field site located in Tsukuba, Japan was sieved (<2mm) and mixed with either N-K (low P) or N-P-K (high P) fertilizer (22.2g fertilizer in 100 l volume of sieved soil containing either 3.33g N and 3.33g K; or 3.33g N, 3.33g P, and 3.33g K, respectively), and used to fill Plexiglas rhizoboxes (30cm × 30cm × 3cm outer dimensions). Seeds were surface-sterilized as described above and at 2 DAS single seedlings were transferred to rhizoboxes watered to field capacity. Five replicates per genotype of DJ123, Santhi Sufaid, Nerica4, Sadri Tor Misri, and Taichung native were grown in a completely randomized block design. Plants were watered daily for 4 weeks using an equal amount of distilled water per box that was sufficient to re-wet any dried portion of the soil while avoiding anaerobic conditions. Upon harvest the rhizoboxes were opened and excess soil was gently washed off to extract roots that were entirely covered. Roots were stored at 4 °C in 50% ethanol prior to evaluation of root hairs.
Field growth conditions
Upland
Seeds were directly sown in upland fields comprised of the same Andosol soil used in the rhizobox experiments. While the high P field was supplied with N-P-K fertilizer (50-50-50kg ha−1) the low P field received only N-K fertilizer (50-0-50kg ha−1), and had not been fertilized with P for 40 years prior to the study (Pariasca-Tanaka et al., 2009). Plants of the genotypes DJ123, Nerica4, Santhi Sufaid, Mudgo, IR64, and Taichung native were grown in three randomized replicate blocks at both field sites. Samples of intact roots were obtained at 50 and 100 DAS as described recently (Mori et al., 2016), and stored in 50% ethanol at 4 °C prior to root hair evaluation.
For root length determination per plant, root systems were obtained 100 DAS, and washed under running tap water. A portion of the root system was stained with ebony black ink (Dylon, UK) in 30% (w/v) NaCl for 30min at 80 °C followed by washing under running tap water for 5min. Using the WinRhizo software (Regent Instruments Inc., Canada) the roots were scanned and sorted into diameter classes representing main roots (>0.4mm), L-type LRs (0.4 mm–0.15mm), or thin LRs (S-type and second order LRs, <0.15mm).
Lowland/paddy field
Sterilized seeds were transferred to a nursery bed 2 d after sowing. Seedlings of genotypes DJ123, Nerica4, Santhi Sufaid, Sadri Tor Misri, Mudgo, and Taichung native were transplanted at 22 DAS in three replications in a Fluvisol soil-containing paddy field with sufficient nutrition (50-50-50kg ha−1 N-P-K fertilizer) located in Tsukuba, Japan; a P-deficient field location was not available. Intact roots were gently removed from the paddy field 28 d after transplanting (DAT), and kept at 4 °C in 50% ethanol prior to root hair evaluation.
Evaluation of root hair formation
For all experiments the longest main root (seminal or crown root) was harvested, on which two 1cm long sections were analysed: a 1cm section at 5cm distance from the root tip and a 1cm section from the most basal part of the root. In each of these 1cm sections, 3–4 images were randomly taken for the main root and its subordinate L-type lateral roots, S-type lateral roots, and if present, second order lateral roots. An Olympus BX50 (Olympus, Tokyo, Japan) light microscope equipped with a digital camera (DP21, Olympus, Tokyo, Japan) was used to obtain the images. These images were then analysed to determine root hair length and density. The length of five randomly chosen root hairs per image was measured using Image J (US National Institutes of Health, Bethesda, USA). Root hair density was qualitatively detected by assigning a score value per picture, ranging from 0 (no hairs) to 5 (very high hair density). By counting the number of root hairs on images with a typical density score, the corresponding root hair number per mm of root were obtained (Supplementary Data, available at JXB online).
For statistical analyses, the Statistix 9 software (Analytical Software, Tallahassee, USA) was utilized. Comparisons between P treatments – and between root types within growth conditions – were performed using a general analysis of variance (ANOVA). The obtained values for root hair length and density were averaged over all genotypes since the ANOVA P treatment × genotype interaction term was not significant. This allowed us to better highlight effects of growth media and differences among root types.
Synchrotron image capture
A sand-textured Eutric Cambisol soil from Abergwyngreygyn, North Wales, was collected, sieved to <5mm, autoclaved, air-dried for 2 d, and then sieved between bounds of 1680 and 1000 µm. This soil was used to fill seven polypropylene syringe barrels of 1ml volume and 4.2mm internal diameter, fitted to a seed cup with root guides (Supplementary Data), fabricated in ABS plastic using an UP! 3D printer (PP3DP, Beijing, China). Seeds of DJ123 and Sadri Tor Misri were germinated for 2 d in filter paper moistened with distilled water, planted in the seed cups, and grown for 14 d (long-day conditions) in a growth chamber (Fitotron SGR, Weiss-Gallenkamp, Loughborough, UK). The syringe chambers were excised and only those containing roots completely surrounded by soil were used for further analysis. Imaging was conducted at the TOMCAT beamline on the X02DA port of the Swiss Light Source, a third generation synchrotron at the Paul Scherrer Institue, Villigen, Switzerland. An image resolution of 1.62 µm was reached using a 19kV monochromatic beam. An exposure time of 90ms was used to collect 1601 projections over 180degrees, at a scan time of 2.4min per sample. Data were reconstructed to 16-bit volumes using a custom back-projection algorithm implementing a Parzen filter (full file size 2560×2560×2180), and analysed using a multi-pass approach. Soil minerals, pore gas and pore fluid were extracted via a heuristic method using a trainable segmentation plugin in the open-source imaging analysis software FIJI (Image J, US National Institutes of Health, Bethesda, USA) that implements the WEKA classifiers for feature detection, using a range of local statistics as training parameters (Hall et al., 2009). The gas phase was filtered using a 3D median filter at a kernel size of 8 voxels (cubic pixel, the smallest discrete element in the digital X-ray CT volume). This region was used to mask out the root hairs in the pore space, which were then segmented via hysteresis thresholding and skeletonized using FIJI. Statistics were calculated using a 3D skeleton analyser in FIJI using trimming of spurious branches resulting from image noise.
Results
Nutrient solution-grown root types differ in their root hair formation
Rice plants grown in nutrient solution containing 1 µM P (low P) or 100 µM P (high P) produced root hairs on main roots (seminal or crown roots), the large L-type lateral roots (L-type LRs), and the small S-type lateral roots (S-type LRs), but not on second order LRs (Fig. 1). Growth in P-deficient nutrient solution led to the formation of longer root hairs on main and L-type LRs, but not on S-type LRs (Fig. 1A). For both P treatments, the observed root hair length on LRs was a third the length of root hairs on the main roots. Root hair density differences were even more pronounced between the root types (Fig. 1B). Where main roots formed at least 50 hairs per mm, L-type LRs only had 5 hairs per mm in high P and 10 hairs per mm in low P conditions. Many S-type LRs did not form any root hairs – leading to an extremely low density value – but significantly more hairs were found under low P conditions. Surprisingly, while many second order LRs were found, none of them formed root hairs under either P treatment.
Root hair growth in nutrient solution. (A) Root hair length determination in nutrient solution including 100 µM P supply (high P, black bars), or 1 µM P (low P, white bars). (B) Root hair density in 100 µM P nutrient solution (high P, black bars), or 1 µM P (low P, grey bars). Depicted are means of ten genotypes, each grown in five replicates, ±standard error (n=50). P-value (***, ≤0.001) represents significant differences between P treatments. Bar indicates Least Significant Difference (LSD) between root types.
Roots grown in rhizoboxes containing P-fixing Andosol soil with (high P) or without (low P) application of P fertilizer showed clear differences in root hair formation as compared to the roots grown in nutrient solution (Fig. 2). Within root types, little difference in root hair length was observed between P treatments, but on average root hairs were 10% longer under low P conditions (Fig. 2A). In contrast to these small differences between P treatments, root hair length showed far greater differences between root types, with main roots producing 40%, 60%, and 80% longer hairs than L-type, S-type, and second order LRs, respectively. Root hair density on the different root types was similar, but significantly more hairs were found on plants grown in the high P condition (Fig. 2B).
Root hair formation in soil-grown greenhouse conditions. (A) Root hair length of plants grown in P fertilized (high P, black bars), or unfertilized (low P, white bars) low P soil in rhizoboxes. (B) Root hair density in P fertilized (high P, black bars), or unfertilized (low P, grey bars) soil. Means of five genotypes, five replicates each, ±standard error (n=25) are shown. P-value (*, ≤0.05; **, ≤0.01; ***, ≤0.001) represents significant differences between P treatments. Bar indicates Least Significant Difference (LSD) between root types.
Comparing nutrient solution-grown and rhizobox-grown roots, a clear difference between main roots and their LRs could be observed. While root hair length on LRs was similar, nutrient solution-grown main roots formed hairs twice the length of rhizobox-grown main roots. Also, main roots developed a similar root hair density under both conditions, but LRs had many more hairs when grown in rhizoboxes. Interestingly, while root hairs were absent on nutrient solution-grown second order LRs, they were abundant in rhizobox-grown second order LRs (Fig. 2B).
In situ imaging of root hairs in intact soil compared to conventional preparation of soil-grown roots
Roots sampled from upland field sites containing highly P-fixing Andosol soil had similar root hair length and density values to rhizobox-grown roots (Fig. 3). Root hair length did not differ between P fertilized (high P) and unfertilized (low P) soil, however main roots formed 36% longer hairs compared to L-type LRs, with this difference increasing to 50% for S-type LRs, and 55% for second order LRs (Fig. 3A). All root types exhibited decreased root hair density when facing P deficiency; the greatest difference was found on main roots and the least on second order LRs (Fig. 3B).
Root hair development in upland field conditions. (A) Root hair length determination of P fertilized (high P, black bars), or unfertilized (low P, white bars) upland field site. (B) Root hair density in the P fertilized upland field soil (high P, black bars), or P unfertilized low P soil (low P, grey bars). Depicted are means of six genotypes, each grown in three replications, ±standard error (n=18). Significant differences between P treatments are represented in P-value (*, ≤0.05; **, ≤0.01). Bar indicates Least Significant Difference (LSD) between root types.
Since root hair lengths in both rhizobox-grown and upland field-grown roots were not even half those of the respective values measured in nutrient solution, the question arises as to whether these unicellular structures are damaged during sample preparation. However, microscopic analyses revealed mostly intact hairs on soil-grown roots (Supplementary Data). Still, it is possible that the intact root hairs observed in microscope images are those remaining after root excavation and washing, while longer hairs might have been completely removed during this process. To follow up on this possibility, in situ images depicting a section of intact main root growing in soil were produced using synchrotron image capture (Fig. 4). Clearly visible is the locality of root hair formation along the epidermis (shown in green) with patches completely lacking hairs where sand grains (in silver-white) – which cannot be pierced by the hairs – were attached, or very close to the root surface. Clay particles (brown), and air pockets (black) were penetrated by root hairs. The resolution allowed identification of single root hairs over several cm of a main root completely surrounded by soil particles. Root imaging of a 180° angle allowed detection of a total of 88% root epidermis via this method. The observed 243 root hairs had a maximum length of 395 µm measured if extending into soil pore space (in black) in between soil particles, and an average length of 122 µm ±78 (SD), which is in a similar range as the conventionally prepared soil samples from rhizobox and upland field experiments. A root hair density of 47mm−2 was detected via this method, similar to the previously detected density values in rhizobox-, and upland field-grown root hairs.
Pseudo-coloured image of a soil-grown seminal root of the genotype DJ123 generated by a synchrotron-based method. The accessible surface of the root and root hairs are depicted in green; portions of the root surface where soil and fluid phases were coincident are marked in red, with measurement not being possible over this region. Soil/clay particles are shown in brown, while sand grains are visualized in silver grey. A section of 5.2mm−2 visible root surface is shown in this image. Scale bar, 500 µm.
Root hair formation in lowland field differs from that under other growth conditions
As root hair formation on nutrient solution-grown roots differed greatly from upland and rhizobox-grown roots, plants grown in flooded lowland fields were investigated for similar characteristics. Only a sufficiently fertilized lowland field was available, thus we compared these results to those for plants grown in a high P nutrient solution.
Root hair formation was detected on all tested root types, and root hair lengths of up to 100 µm were found (Fig. 5). These root hair lengths per root type were lower when compared to those of nutrient solution-grown roots (Fig. 5A). The only exception were the second order LRs, which formed root hairs, as opposed to the nutrient solution-grown second order LRs, which did not. While root hair length was similar among LR types, main roots in nutrient solution developed hairs more than twice the length of those grown in lowland soil. In addition, root hair density of lowland-grown root types was very low, decreasing respectively from main, to L-type, S-type, and second order LRs (Fig. 5B). While root hair densities were similar on lowland-grown and to nutrient solution-grown L-type and S-type LRs, differences were found for main roots and second order LRs. Lowland-grown main roots formed only a fifth the number of root hairs, and second order LRs differed generally by forming root hairs compared to the nutrient solution-grown second order LRs.
Root hair emergence in lowland paddy field. (A) Root hair length in lowland field condition including sufficient P supply (high P). (B) Root hair density in sufficiently supplied lowland field (high P). Depicted are means of six genotypes, each in three replications, ±standard error (n=18). Bar indicates Least Significant Difference (LSD) between root types.
To summarize differences among the growth conditions, root hair lengths and densities for all treatments were normalized to the high P upland field per root type (Table 1). The effect of growth medium was larger than the effect of P treatment. While upland and rhizobox-grown main roots formed root hairs of similar length, nutrient solution-grown hairs were twice the length, whereas lowland-grown hairs were half the length of the high P upland field-grown hairs. A similar distribution was found for root hair density on main roots, but upland and rhizobox-grown roots produced more hairs in the high P condition, compared to the low P condition.
Root hair formation under different growth conditions
| Growth condition | Main roots | L-type LRs | S-type LRs | Second order LRs |
|---|---|---|---|---|
| Normalized root hair length [µm] a | ||||
| Nutrient solution, low P | 268 | 145 | 108 | 0 |
| Nutrient solution, high P | 197 | 81 | 87 | 0 |
| Rhizobox, low P | 118 | 132 | 119 | 130 |
| Rhizobox, high P | 112 | 115 | 108 | 104 |
| Upland field, low P | 97 | 96 | 101 | 89 |
| Upland field, high P | 100 | 100 | 100 | 100 |
| Lowland field, high P | 57 | 34 | 28 | 19 |
| Normalized root hair density [1/mm] a | ||||
| Nutrient solution, low P | 182 | 39 | 2 | 0 |
| Nutrient solution, high P | 173 | 17 | 1 | 0 |
| Rhizobox, low P | 94 | 112 | 100 | 157 |
| Rhizobox, high P | 122 | 128 | 158 | 236 |
| Upland field, low P | 89 | 73 | 77 | 64 |
| Upland field, high P | 100 | 100 | 100 | 100 |
| Lowland field, high P | 37 | 24 | 21 | 20 |
| Growth condition | Main roots | L-type LRs | S-type LRs | Second order LRs |
|---|---|---|---|---|
| Normalized root hair length [µm] a | ||||
| Nutrient solution, low P | 268 | 145 | 108 | 0 |
| Nutrient solution, high P | 197 | 81 | 87 | 0 |
| Rhizobox, low P | 118 | 132 | 119 | 130 |
| Rhizobox, high P | 112 | 115 | 108 | 104 |
| Upland field, low P | 97 | 96 | 101 | 89 |
| Upland field, high P | 100 | 100 | 100 | 100 |
| Lowland field, high P | 57 | 34 | 28 | 19 |
| Normalized root hair density [1/mm] a | ||||
| Nutrient solution, low P | 182 | 39 | 2 | 0 |
| Nutrient solution, high P | 173 | 17 | 1 | 0 |
| Rhizobox, low P | 94 | 112 | 100 | 157 |
| Rhizobox, high P | 122 | 128 | 158 | 236 |
| Upland field, low P | 89 | 73 | 77 | 64 |
| Upland field, high P | 100 | 100 | 100 | 100 |
| Lowland field, high P | 37 | 24 | 21 | 20 |
Data normalized to upland field (high P) per root type.
Root hair formation under different growth conditions
| Growth condition | Main roots | L-type LRs | S-type LRs | Second order LRs |
|---|---|---|---|---|
| Normalized root hair length [µm] a | ||||
| Nutrient solution, low P | 268 | 145 | 108 | 0 |
| Nutrient solution, high P | 197 | 81 | 87 | 0 |
| Rhizobox, low P | 118 | 132 | 119 | 130 |
| Rhizobox, high P | 112 | 115 | 108 | 104 |
| Upland field, low P | 97 | 96 | 101 | 89 |
| Upland field, high P | 100 | 100 | 100 | 100 |
| Lowland field, high P | 57 | 34 | 28 | 19 |
| Normalized root hair density [1/mm] a | ||||
| Nutrient solution, low P | 182 | 39 | 2 | 0 |
| Nutrient solution, high P | 173 | 17 | 1 | 0 |
| Rhizobox, low P | 94 | 112 | 100 | 157 |
| Rhizobox, high P | 122 | 128 | 158 | 236 |
| Upland field, low P | 89 | 73 | 77 | 64 |
| Upland field, high P | 100 | 100 | 100 | 100 |
| Lowland field, high P | 37 | 24 | 21 | 20 |
| Growth condition | Main roots | L-type LRs | S-type LRs | Second order LRs |
|---|---|---|---|---|
| Normalized root hair length [µm] a | ||||
| Nutrient solution, low P | 268 | 145 | 108 | 0 |
| Nutrient solution, high P | 197 | 81 | 87 | 0 |
| Rhizobox, low P | 118 | 132 | 119 | 130 |
| Rhizobox, high P | 112 | 115 | 108 | 104 |
| Upland field, low P | 97 | 96 | 101 | 89 |
| Upland field, high P | 100 | 100 | 100 | 100 |
| Lowland field, high P | 57 | 34 | 28 | 19 |
| Normalized root hair density [1/mm] a | ||||
| Nutrient solution, low P | 182 | 39 | 2 | 0 |
| Nutrient solution, high P | 173 | 17 | 1 | 0 |
| Rhizobox, low P | 94 | 112 | 100 | 157 |
| Rhizobox, high P | 122 | 128 | 158 | 236 |
| Upland field, low P | 89 | 73 | 77 | 64 |
| Upland field, high P | 100 | 100 | 100 | 100 |
| Lowland field, high P | 37 | 24 | 21 | 20 |
Data normalized to upland field (high P) per root type.
Interestingly, the growth medium effect was small if only L-type and S-type LR types were considered. On L-type LRs, the longest hairs were formed in P deficient nutrient solution and in rhizoboxes, while high P nutrient solution and the lowland field produced the shortest hairs. L-type LRs developed the fewest hairs in lowland soil and both nutrient solution conditions, and the most in both rhizobox P treatments. S-type LRs, on the other hand, differed less in hair length, only being substantially shorter in lowland-grown plants, but their root hair density was extremely reduced in nutrient solution-grown roots, even below the values in the lowland field. While plants in all other growth conditions had hair-forming second order LRs, hairs were not formed on these roots in nutrient solution. Again, root hair length and number were very low in lowland-grown roots.
Root type-specific root hair influence on the root system
Instead of focusing on root hair length and density separately, a combination of these data was devised to improve understanding of developmental differences among experimental designs. The root hair factor (RHF), derived by multiplying length and number of root hairs per root type, was introduced as an approximation of total root hair influence (Fig. 6A). While total RHFs of all growth conditions did vary between 9.1 and 25.0, the distribution among root types differed to a much greater extend. Nutrient solution-grown main roots had RHF values of 16.0 and 23.1 for high and low P samples, contributing to 97% and 92% of total RHF, respectively. In contrast, upland and rhizobox-grown main roots only attained RHFs between 4.0 and 6.4, corresponding to between 33% and 44% of total RHF, respectively. Lowland-grown main roots had a calculated RHF of 1.0 or 68% of total RHF. All lateral roots combined contributed to <10% of the total RHF for nutrient solution-grown plants. In contrast, upland and rhizobox-grown LRs accounted for the majority of total RHF with decreasing percentages from L-type, over S-type, to second order LRs. Lowland-grown LRs contributed to a third of total RHF, also with decreasing amounts from L-type, S-type, to second order LRs.
![Root type contribution to RHF and root length. (A) Relative root hair factor (RHF), obtained by multiplication of root hair length (mm) with root hair number [1mm−1] per root type for each tested growth condition. (B) Relative root length per root type in the whole root system of upland-grown plants. Absolute values are shown in the bars per root type, and total absolute RHF and root length over all root types are depicted on top of the graph.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/jxb/67/12/10.1093_jxb_erw115/2/m_exbotj_erw115_f0006.jpeg?Expires=1748013790&Signature=FmzvPKrqyKN3GrX7plvia9t~xO5MecuNfGiHT4U1phd--ds7lUqlNqx3v28ONPVe4THmEDVAecflO7spaIw6V4XwsBDeFXh3ZNDaEPuJTl14kWaYascsR4CLpmKettMjizbtG~-GTID4KCS5pTdXSg~YEmYEqnn2yfhYZEoXE7ijYwzbKj2AbIceT6pE24~bbBHrSHqaIK6BDZoUIscmZrHhwxWmcwj-uXoJSAbXgOBxGaFPL9Q5JNHc99gBzcVTB~Ni8G2zGTBPtbL9k4tptcDJqQMYWlLmlo9ScMqUGvFIRMbnMXgIFxVRuNLjt2dtjjBiio7c3LxHbO0dOQ8IQw__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Root type contribution to RHF and root length. (A) Relative root hair factor (RHF), obtained by multiplication of root hair length (mm) with root hair number [1mm−1] per root type for each tested growth condition. (B) Relative root length per root type in the whole root system of upland-grown plants. Absolute values are shown in the bars per root type, and total absolute RHF and root length over all root types are depicted on top of the graph.
Upland field-grown plants were used for root scanning to obtain the total root length of the root system (Fig. 6B). While plants in the high P field had nearly twice the total root length of the low P field, the distribution among root types was similar. Only 10–15% of total root length was made up from main roots, another 15% of L-type LRs, and the remaining majority of 70–75% of the root system was comprised of S-type or second order LRs, which were analysed in combination by this scanning method.
Discussion
Rice root hairs in soil are short and show little response to P treatment
Main roots as well as all LR types analysed in this study developed short root hairs – <200 µm – when grown in upland soil, independent of growth in a field (Fig. 3) or in rhizoboxes (Fig. 2), and showed little to no response to P level. By contrast, roots grown in low P nutrient solution developed longer root hairs than those grown in high P conditions (Fig. 1), as was expected from previous studies on different plant species (Foehse and Jungk, 1983; Bates and Lynch, 1996). Why did we find little to no difference in soil-grown plant root hair length in response to P level when testing the same genotypes of rice at similar or even the same age in all growth conditions? Synchrotron-based imaging (Fig. 4), which provided the first in situ images of rice root hairs, indicated that soil particles like sand grains possibly limited root hair formation when in direct contact with the root epidermis and root hair extension upon contact with growing root hairs. Despite considerable differences in soil and growth conditions between samples from field, rhizobox, and synchrotron studies, the root hair length and density values obtained by the synchrotron method were similar to those detected for soil-grown roots by the conventional analysis method. Thus, as root hairs grow unidirectionally, sand grains and other hard particles seem to represent impenetrable barriers that limit the maximum length of root hairs under natural conditions. While we identified an average root hair length of 122 µm, a three times greater maximum length of 395 µm was detected by synchrotron imaging, highlighting the importance of distinguishing between maximum and average root hair length in soil. In the only other in situ imaging study of root hairs published, a similar pattern was found, with average wheat root hairs of 300–400 µm, but individual hairs growing twice as long (Keyes et al., 2013).
Furthermore, higher soil strength in the low P upland field compared to the high P field might have decreased root hair length by presenting a higher penetration resistance, as barley plants were shown to form shorter root hairs when grown in soils of higher strength (Haling et al., 2013). In accordance with these results, plants grown in the same, but sieved, upland soil in rhizoboxes produced longer hairs compared to those grown in the more compact soil of the upland fields, but only the mean root hair length of all root types was greater for low P compared to high P soil. Interestingly, a study in maize found increased soil moisture had a negative impact on root hair growth that was greater than the impact of soil P status (Mackay and Barber, 1985).
In contrast to the small difference identified in soil-grown roots (Figs 2, 3), low P availability led to an increase of 40% (main roots) to 80% (L-type LRs) in root hair length on nutrient solution-grown roots (Fig. 1). These values are far lower compared to studies on other species grown in nutrient solution, such as tomato with 200%, rape and spinach with 300% (Foehse and Jungk 1983), and Arabidopsis, with up to 500% longer root hairs in similar P-limiting conditions (Bates and Lynch, 1996). Surprisingly, a recent study on 166 Arabidopsis accessions revealed no correlation of local low P availability with root hair length (Stetter et al., 2015). These diverse results indicate that although in some species increases in root hair length are correlated with low P status, this does not hold true for other species or even for other varieties. In line with this, barley varieties producing longer root hairs maintain high yield in low P soil compared to shorter root hair varieties, which have decreased yield in low P compared to high P soil (Gahoonia and Nielsen, 2004), but in a later study it was concluded that the presence of barley root hairs alone, and not their length, was important to maintaining yield in P-limited conditions (Brown et al., 2012).
It is noteworthy that while soil properties can explain the lack of differences in root hair length in our study, they cannot completely account for the discrepancy found in root hair density in low P compared to high P treatments. Soil compactness and the resulting penetration resistance, and hard particles such as sand grains can explain the lower densities found in upland soil-grown compared to nutrient solution-grown plants. But in low P nutrient solution-grown plants an increase in root hair density was detected compared to the high P condition, while upland field- and rhizobox-grown roots exhibited a decrease in root hair number in the P limiting condition. It is possible, but unlikely, that the low P upland soil used in this study contained considerably more sand grains than the high P soil, but that might also have decreased root hair length below the high P field levels.
Rice root hair growth in nutrient solution is not representative of its development under natural conditions
The comparison of growth conditions tested in this study revealed a clear discrepancy between the artificial growth conditions in nutrient solution and all soil-based environments (Table 1, Fig. 6A). As nutrient solution-grown main roots developed much longer root hairs than their soil-grown counterparts it can be argued that a lack of physical penetration resistance played a part, but experiments in agar have yielded similar results to those in nutrient solution (unpublished data). Even growth in flooded lowland fields differed greatly from growth in nutrient solution, indicating that higher water content and reduced physical resistance are not the only factors influencing root hair formation. Previously, lowland-grown rice roots were shown to form up to 200 µm long root hairs (Kawata and Ishihara, 1961), a result which corresponds to our findings. Though a different soil was used for the synchrotron-based detection of root hairs (Fig. 4), similarly long and dense hairs compared to upland and rhizobox-grown conditions were identified, emphasizing the clear developmental differences of root hairs in soil when compared to the artificial conditions in nutrient solutions. Interestingly, in maize it was shown that drying of soil increased formation of longer hairs while wetter soil induced shorter root hairs (MacKay and Barber, 1987), which is in line with our findings that higher soil moisture in the flooded lowlands (Fig. 5) reduced root hair length and density compared to upland field conditions (Fig. 3). The very low, or even absence of, root hair density on S-type and second order LRs grown in nutrient solution (Fig. 1B) is the most obvious difference to all soil-based methods used here. This might indicate a rice-specific root architectural adaptation which will be discussed in the third part of the discussion. In conclusion, growth in nutrient solution is not representative of growth in natural soils, but on main roots, it might indicate the potential of maximal root hair length under minimal penetration resistance.
Plant species differ in root hair initiation patterns (Clowes, 2000) as well as in potential length and density. For instance, maize was shown to form ~1000 µm long root hairs when germinated in filter paper (Nestler et al., 2014), but much shorter hairs (200–400 µm) were detected under different soil conditions (Barber and Mackay, 1986). In wheat, hairs up to 1000 µm were found (Delhaize et al., 2015), and barley had 600–800 µm long hairs when grown in soil (Brown et al., 2012). Our observations of 100–200 µm long root hairs in soil are lower compared to other cereals, however they are comparable in length to the 200–300 µm hairs characteristic of peanut (Arachis hypogaea) detected in the same soil (Wissuwa and Ae, 2001), and the 100 µm hairs measured in soil grown samples of Oryza sativa cv. Nipponbare (Wissuwa, 2003).
Root hairs in context of a root system
Our study provides the first data on differences in root hair development between various root types in rice. Strikingly, when grown in nutrient solution, root hair formation on the different root types exhibits very large variation, with the main roots giving rise to much longer root hairs at a much higher density than their LRs (Fig. 1). Maize grown in water-saturated filter paper also formed slightly shorter root hairs on LRs compared to those on the main roots (Nestler et al., 2014), and when grown in nutrient solution, maize LRs had slightly longer root hairs than our findings (Gaudin et al., 2011). The most striking distinction of growth in nutrient solution compared to growth in soil was the absence of root hairs on second order LRs, and the near-absence on S-type LRs (Fig. 1A). A striking 70 to nearly 80% of total root length of the rice root system comprises thin LRs, S-type as well as second order LRs <150 µm diameter (Fig. 6B). Thus, the conventional measurement of hair traits using main roots, which represent only a minority of the root system, may lead to hair length and density parameters being overestimated, if data from nutrient solution-grown plants is extrapolated to the field.
Compared to the big differences among root types detected for rice grown in nutrient solution, all soil-grown approaches yielded smaller differences (Fig. 6A). Interestingly, the root hair length gradually decreased from L-type, over S-type, to second order LRs, a pattern that was also found, but less pronounced, for root hair density. The high contribution of thin LRs, S-type and second order LRs, to the total root system length, and the presence of root hairs on these short and fine LRs might be a unique feature of the rice root system. Little is known about the contribution of S-type LRs to root system size in other cereals. LRs of varying diameter were detected in maize (Cahn et al., 1989), but no distinct types are defined. In contrast, Yamauchi et al. (1987) suggested that most grasses, including maize, barley, and wheat, possess distinct L- and S-type LRs. The root system of Brachypodium distachyon contains seven histologically different lateral root types (Watt et al., 2009).
Compared to the long L-type LRs (1–2cm−1 main root) many more short S-type LRs are formed (10–20cm−1 main root, Supplementary Data). To overcome a potential detrimental competition for nutrient uptake between the LRs and their root hairs, formation of rather short root hairs is sufficient for these distances. The features of thin LRs, S-type LRs and second order LRs – short, non-branching, thin, abundant – in the rice root system can also be interpreted functionally as ‘super root hairs’, or in-between lateral roots and root hairs; increasing the root surface area and nutrient uptake while at the same time being resource cost efficient compared to the large L-type LRs. This might indicate a general developmental difference in LRs among crop species, as barley, wheat, and maize form larger, more voluminous, but coarser root systems compared to rice (Yamauchi et al., 1987). Formation of long root hairs can aid exploration of a large soil volume in these rather coarse-rooted cereals, but we hypothesize them to be less beneficial in rice, potentially leading to detrimental competition between LRs, while shorter root hairs are sufficient. Further studies are needed to fully explore the 3D root geometry in rice and to derive an optimum root hair length that would avoid overlapping zones of P depletion.
Typically, root hair observations are made on one root type, often a main root (seminal or crown root) simplifying the root system by assuming all root types behave similarly. However, P uptake optimizations should occur in context of the entire root system including all lateral root types and 3D geometry. The optimal experimental setting would include synchrotron-based detection of root hairs in intact soil on all root types over representative portions of the entire root system. Currently, the technique and computing power allow only small sample sizes of both root area and replication number. Laborious soil-based experiments therefore remain necessary to obtain a representative picture of root hair development across different root types.
Conclusion
The majority of the mature rice root system is comprised of lateral, in particular fine lateral, roots. To properly evaluate root hair development and their potential contribution to nutrient uptake on a root system scale, root hair properties should be evaluated on these root types. The present study indicated that this can only be achieved through evaluation of root hairs on roots of soil-grown plants. Roots grown in nutrient solution hardly developed any root hairs on the fine S-type and second order LRs, and highly overestimated root hair length and density on the main roots, thus providing a rather distorted image of root hair development. Nevertheless, as soil-based studies of root traits are much more laborious, artificial settings such as growth in nutrient solution could still provide useful estimations of potential maximum root hair length in the screening of large populations. However, such screenings would require confirmation in soil across root types.
Acknowledgments
We would like to thank Pu Chen, Takuya Fukuda, and Juan Pariasca-Tanaka for assistance with the experimental work. This study was partly funded by a postdoctoral fellowship awarded to JN by the Japanese Society for Promotion of Science (JSPS) and by a grant from the Japan Science and Technology Agency (JST) to facilitate collaboration with the EURoot research consortium.
References
Author notes
Editor: Greg Rebetzke, CSIRO, Plant Industries
Comments