# Source paper archive

- Source title: Allometric analysis reveals distinct anatomical patterning and genetic architecture across wheat nodal roots
- Author(s): Tawale, A. B., Schierenbeck, M., Jones, D. H., Fusi, R., Lopez-Valdivia, I., Wolf, A., Linow, P., Sandoni, D., D'Oria, A., Arkoun, M., Dietz, J. I., Campos, P. E., Boerner, A., Golan, G., Schneider, H.
- Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.16.766962v1
- PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.16.766962v1.full.pdf
- PDF SHA-256: c82b977d0e7b130402d89dd0bbadb07bb2a01ca8fa6c6992df6269a9761adb9a
- Converter: pymupdf4llm 0.0.27
- Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version.

---

&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)

# 1 Allometric analysis reveals distinct anatomical patterning and 2 genetic architecture across wheat nodal roots

## 3 Akshay B. Tawale ¹*, Matías Schierenbeck ¹*, Dylan H. Jones ¹, Riccardo Fusi ¹, Ivan Lopez- 4 Valdivia ¹, Annegret Wolf ¹, Petra Linow ¹, Daniele Sandoni ¹, Aurélien D'Oria &#91;2&#93;, Mustapha

5 Arkoun &#91;2&#93;, Juan I. Dietz &#91;3,4&#93;, Pablo E. Campos &#91;5&#93;, Andreas Börner &#91;1&#93;, Guy Golan &#91;6&#93;, Hannah M.
## 6 Schneider &#91;1,7&#93; *


7 Affiliations:

## 8 ¹ Leibniz Institute of Plant Genetics and Crop Plant Research (IPK), Corrensstr. 3, OT

9 Gatersleben, Seeland, D-06466, Germany
2
10 Centre Mondial de l’Innovation Roullier, Saint-Malo, France

3
11 National Council of Scientific and Technical Research, Buenos Aires (CONICET),

12 Argentina
4
13 Cereals, Faculty of Agricultural and Forestry Sciences, National University of La Plata, La

14 Plata, Argentina
5
15 Bordenave Agricultural Experimental Station CC 44, National Institute of Agricultural

16 Technology (INTA), Buenos Aires, Argentina
6
17 The Institute of Plant Sciences and Genetics in Agriculture, The Robert H. Smith Faculty of

18 Agriculture, Food and Environment, The Hebrew University of Jerusalem, Rehovot 7610001,

19 Israel

7
20 Division of Root Sciences, Department of Crop Science, Georg-August-University

21 Göttingen, Göttingen, Germany
## 22 * Authors for correspondence: Akshay B. Tawale (tawale@ipk-gatersleben.de), Matias

23 Schierenbeck (schierenbeck@ipk-gatersleben.de) and Hannah M. Schneider
24 (schneiderh@ipk-gatersleben.de)

25


1


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


26 **Author contributions**

27 ABT and HMS conceived the study. ABT, MS, and HMS designed the study. Plant growth
28 and wet-lab work were performed by ABT, MS, ILV, DHJ, AW, DS, and PL. Dry-lab
29 methodology was performed by ABT, AW, and PL. ABT, HMS, MS, and RF performed data
30 analyses. Funding was acquired by HMS, AD, MA and MS. ABT, MS, and HMS drafted the
31 manuscript. The wheat diversity panel was generated and made available by MS, JD, PC, and
32 AB. AD, MA, and GG advised on conceptualisation. All authors contributed substantially to

33 the revisions.


34 **Funding**
35 This study was kindly supported by CMI Roullier, France, and open-access funding is
36 enabled by the project RootOptiGrowth. The Alexander von Humboldt Foundation (AvH)
37 provided the funding for genotyping the wheat diversity (TangoWheat) panel.


38 **Declaration of generative AI and AI-assisted technologies in the manuscript**
39 **preparation process**
40 During the preparation of the work, the authors used Claude Sonnet 5 (Anthropic), via the
41 ZGPK platform (IPK Gatersleben), to assist with English phrasing and R code editing. The
42 authors reviewed all AI-assisted content and take full responsibility for the final article.


43 **Conflicts of interest**

44 The authors declare no conflicts of interest.


45 **Code and data availability**
46 Data are available in the article's supplementary material.


47 **Acknowledgments**
48 We thank Corinna Trautewig, Avani Vallayil, Darshan Damodaran, Pauline Hampe, Maria
49 Schoen, Zhongyang Wang, Luisa Meier, Luisa Wolfgramm, Rashmi Aryal, Osalumen
50 Philemon Irusota, Wirassaya Tobdan, Gopika Shaju, and Misbah Chaudhry for their
51 assistance with plant evaluations and sample collections from the field and greenhouse. We
52 thank Peter Schreiber, Enk Geyer, Gerald Fiedler, Oliver Schmidt, Stefanie Fischer, Kerstin
53 Jacobs, Anke Mueller, and Kathrin Tiemann for their assistance with greenhouse and field
54 experiment management.


55 **Abstract**


56 Root anatomical traits influence resource acquisition, metabolic costs, hydraulic conductance,


57 and stress adaptation in cereal crops. Although classified as a single root class, nodal roots


58 may differ anatomically according to their position along the culm. As shown in maize, this


59 variation is particularly notable in root cross-sectional area (RxA). However, it remains


60 unclear whether these differences reflect developmental differentiation or simply variation


2


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


61 associated with RxA. We quantified root anatomical traits across four main-culm nodes


62 (basal to apical) in 190 hexaploid wheat genotypes grown under field and greenhouse


63 conditions. To distinguish RxA-dependent from RxA-independent anatomical variation, we


64 applied allometric size correction, adjusting traits to a reference RxA, and conducted parallel


65 genome-wide association studies (GWAS) using raw and size-corrected traits. Raw


66 measurements revealed increases in RxA and cortical area from basal to apical nodes,


67 accompanied by decreases in metaxylem area. After correction to a common reference RxA


68 within each environment, anatomical differences among nodal positions persisted, with roots


69 emerging from more apical nodes exhibiting smaller stele and metaxylem areas. Differences


70 in cortical area were primarily associated with greater cortical cell file layer number rather


71 than increased mean cortical cell diameter. GWAS identified largely distinct genetic


72 associations for raw and RxA-corrected anatomical traits, varying across nodal positions and


73 environments. Together, these findings demonstrate that anatomical variation across the


74 wheat nodal root system cannot be explained by RxA alone. Our results highlight nodal


75 position and allometric scaling as important considerations for anatomical phenotyping and


76 distinguishing genetic associations with overall root cross-sectional area from those with


77 tissue organisation.


78 **Keywords:** Anatomy, Allometry, Nodal Roots, Wheat


79 **Abbreviations** : Nodal roots NR, Main culm nodal (N) root positions (basal to apical) (Node


80 1 to Node 4; N1 to N4, genome-wide association studies GWAS, Root cross-sectional area


81 RxA, Stele cross-sectional area SxA, Cortex area CA, Cortex to stele ratio CSR, Cortical cell


82 file layer number CCFN, Mean cortical cell diameter mCCD, Metaxylem vessel number


83 MXN, Total metaxylem area tMXA, Mean metaxylem area mMXA, Metaxylem to stele ratio


84 MSR


3


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


85 **Graphical abstract**


86


87 **1.** **Introduction**


88 Wheat ( _Triticum aestivum_ ) is one of the most important cereal crops, providing


89 approximately 20% of global dietary calories ( _1_ ). It is the third most widely produced cereal


90 after maize and rice, and remains a major component of human diets worldwide. Maintaining


91 wheat productivity under changing climates and increasing resource limitations will require a


92 better understanding of the traits that determine resource acquisition and stress adaptation ( _2_ ).


93 Roots are central to these processes because they acquire water and nutrients, anchor the


94 plant to the soil, and mediate responses to edaphic stress ( _3_ ). Despite these functions, root


95 traits remain comparatively underexplored, in part because techniques for phenotyping


96 belowground organs are relatively underdeveloped compared to aboveground tissues, and


97 limited sample accessibility in field conditions makes root studies especially challenging ( _4_ ).


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


98 Wheat, like most grasses, develops a fibrous root system composed of roots with distinct


99 developmental origins: embryonic (also called seminal) roots that develop from the embryo


100 and post-embryonic (or nodal) roots that emerge from culm nodes. Roots emerging together


101 from a single culm node are referred to collectively as nodal root cohorts. Both root types


102 subsequently produce root-borne lateral roots ( _5_, _3_ ). In wheat and other cereals, post

103 embryonic nodal roots develop from basal nodes of the main culm and tillers ( _6_ – _8_ ). These


104 basal nodes remain tightly compressed due to minimal internode elongation, with successive


105 phytomer lengths increasing non-uniformly along the main culm axis ( _9_ ). However, despite


106 originating in close physical proximity along the culm, successive nodal root cohorts appear


107 sequentially throughout plant development, driven by thermal time accumulation ( _10_ ).


108 While the embryonic root system dominates early development in wheat and barley


109 ( _Hordeum vulgare_ ) ( _11_ – _13_ ), nodal roots can make an increasingly large contribution to water


110 and nitrogen acquisition as plant matures ( _14_ – _18_ ). For example, a hydroponic study of wheat,


111 barley, and rye ( _Secale cereale_ ) found that between anthesis and physiological maturity,


112 nodal roots accounted for approximately 75% of root water uptake, with seminal roots


113 contributing the remaining 25% ( _19_ ). Under hypoxic or waterlogged conditions, wheat nodal


114 roots can maintain greater nutrient uptake and remain functional longer than seminal roots


115 ( _20_ ). Together, these findings highlight the increasing importance of nodal roots as plants


116 mature and their critical role in sustaining resource acquisition under both favourable and


117 stressful conditions ( _15_, _17_, _18_, _21_ ).


118 While the contribution of nodal roots to resource acquisition changes as plants mature, their


119 functional capacity is also determined by their internal anatomical organisation. Root


120 anatomical traits influence the metabolic cost of soil exploration, mechanical properties, and


121 the transport of water and nutrients ( _22_, _23_ ). Plants can reduce the metabolic cost of root


122 construction and maintenance through anatomical phenotypes, including reduced cortical


123 area, increased cortical cell size, fewer cortical cell layers, and greater aerenchyma formation


124 in the cortex. These traits can reduce respiratory costs and enable greater soil exploration


125 ( _22_ ). Increased root diameter can prevent buckling, facilitate penetration in hard soils, and


126 support plant anchorage ( _24_ – _26_ ). Root anatomy also regulates hydraulic function: fewer and


127 narrower metaxylem vessels can reduce axial hydraulic conductance, potentially limiting


128 early-season water use and conserving soil water for later reproductive development, and


129 tissues in the cortex, including aerenchyma, can influence radial transport of water and


5


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


130 nutrients ( _27_, _28_ ). Thus, variation in root anatomy can have direct consequences for both


131 resource acquisition and whole-plant performance ( _29_ – _31_ ).


132 Importantly, root anatomy is not uniform across the cereal root system. Anatomical traits


133 differ between embryonic and nodal root classes in cereals such as maize, barley, and


134 intermediate wheatgrass ( _Thinopyrum intermedium_ ) ( _21_, _32_, _33_ ). These root classes have


135 distinct developmental origins and can differ in their genetic regulation ( _7_ ). However,


136 anatomical variation may not be limited to differences between embryonic and post

137 embryonic roots. Evidence from maize indicates that nodal roots originating from successive


138 culm nodal positions differ systematically in their anatomy and architecture; in particular,


139 apical nodal roots typically exhibit larger cross-sectional, cortical, and stelar areas than roots


140 from basal nodes ( _33_ – _35_ ).


141 Such variation is particularly relevant because cereals produce nodal roots sequentially from


142 multiple culm nodes as the plant matures; typically, up to nine nodes in maize and three to


143 seven in wheat ( _36_, _37_, _10_ ). Root cohorts emerging from successive culm nodes therefore


144 represent a structured developmental series rather than repeated, functionally equivalent units


145 of a single root class. Similar positional differentiation occurs aboveground, where leaves and


146 internodes associated with successive phytomers differ systematically in their morphology


147 and function, for example, the flag leaf in wheat ( _36_, _38_ ). Thus, culm node position may


148 provide a developmental coordinate associated with anatomical and potentially functional


149 differentiation within the cereal root system. Consequently, treating all nodal roots as a single


150 root class, or assuming that anatomical measurements of early-emerging basal roots represent


151 those produced later in plant development, may obscure developmental variation, particularly


152 when environmental conditions change over time ( _39_ ).


153 A major challenge, however, is determining whether anatomical differences in root cohorts


154 across nodal positions reflect developmental differentiation or simply differences in root


155 cross-sectional area. Root tissues commonly exhibit allometric relationships, in which the


156 dimensions of individual tissues change predictably as total root cross-sectional area


157 increases ( _34_ ) Thus, without isolating this RxA-dependent scaling, it is not possible to


158 determine whether successive nodal cohorts follow distinct developmental blueprints for


159 tissue organisation or are simply larger versions of the same developmental blueprint.


160 Distinguishing RxA-dependent scaling from RxA-independent developmental differentiation


161 is therefore essential to accurately phenotype variation across the nodal root system.


6


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


162 Resolving this dependence of anatomical traits on RxA is equally important for uncovering


163 the true genetic architecture of root anatomy. Without correction, genetic associations


164 mapped to anatomical traits may partly reflect variation in RxA rather than tissue


165 organisation alone. Furthermore, conventional normalisation approaches, such as simple


166 ratios or linear residual correction, may not adequately account for nonlinear scaling


167 relationships between RxA and individual tissues, potentially introducing statistical artefacts


168 or obscuring biological associations ( _40_ ). Size correction provides an alternative by


169 estimating trait-specific scaling relationships and adjusting anatomical measurements to a


170 common RxA ( _41_ – _44_ ). By applying this framework across genotypes and nodal positions, we


171 can distinguish genetic associations with overall root cross-sectional area from those


172 associated with tissue organisation after accounting for RxA ( _44_ – _46_ ). However, it remains


173 unclear how wheat root anatomy varies across successive nodal positions, how much of this


174 variation is explained by allometric scaling, and whether raw and size-corrected anatomical


175 traits yield distinct genetic architectures.


176 We hypothesised that anatomical variation across successive nodal positions in wheat reflects


177 developmental differentiation, extending beyond mere differences in root cross-sectional


178 area. To test this hypothesis, we characterised root anatomy across four nodal cohorts in a


179 diversity panel of 190 hexaploid wheat genotypes grown under field and greenhouse


180 conditions. Because total root cross-sectional area (RxA) represents the primary physical


181 boundary constraining tissue expansion, we quantified raw anatomical traits and their


182 allometric relationships with RxA, subsequently applying allometric size correction to


183 distinguish RxA-dependent from RxA-independent variation. Throughout this study, “size

184 corrected” traits refer to anatomical measurements adjusted to a common reference RxA. We


185 then conducted parallel genome-wide association studies (GWAS) on both raw and size

186 corrected anatomical traits to identify genetic loci associated with overall anatomical size


187 versus intrinsic tissue organisation. Together, these analyses provide a framework for


188 characterising root anatomical differentiation across nodal positions while explicitly


189 accounting for allometry. Because culm nodal position is inherently associated with


190 differences in root emergence time, root age, and whole-plant developmental stage, we


191 interpret this anatomical variation within a broader developmental context rather than as a


192 rigid, isolated effect of the node itself.


7


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


193 **2.** **Materials and methods**


194 2.1. Plant material


195 A genetic diversity panel of 190 hexaploid winter and spring wheat ( _Triticum aestivum_ L., 2n


196 = 6x = 42) genotypes, designated the Tango-Wheat Panel, was used in this study. This panel


197 was structured to represent modern elite germplasm, primarily adapted to temperate


198 agricultural zones. It comprises 157 commercial cultivars released in Argentina between 2000


199 and 2024, one from Brazil, 25 from Uruguay, two from Germany, and five elite lines from


200 the International Maize and Wheat Improvement Centre (CIMMYT, Mexico). Detailed


201 information, growth habit (days from emergence to flag leaf expansion (GS39) and days from


202 emergence to anthesis initiation (GS61)), and origin for the entire panel are compiled in


203 Table S1.


204 2.2. Field trial


205 The Tango-Wheat panel was grown during the spring season (March 2025 to July 2025) on


206 the experimental field of the Leibniz Institute of Plant Genetics and Crop Plant Research


207 (IPK), Gatersleben, Germany (51°49'22.0" N 11°16'13.2" E). The experimental soil type is


208 Chernozem, characterised by a sandy clay loam texture ( _47_ ). Seeds were sown using an


209 automated experimental seeder at a standardised density of 300 seeds m &#91;−2&#93; . The trial was laid


210 out as a flat-block design with 190 individual genotypes, each allocated to a single plot


211 measuring 1.2 m² (1.2 m length × 1.0 m width) with six longitudinal rows spaced 0.2 m apart.


212 Perimeter buffer plots containing the commercial cultivar 'KWS Quintus' were sown around


213 the entire trial boundary to minimize edge effects.


214 The crop was managed using conventional tillage and standard agronomic practices. Nitrogen


215 fertiliser was applied as urea in three split applications of 60 kg N ha &#91;-&#93; ¹ each: at the onset of


216 the vegetative period (GS13, ( _48_ )), during stem elongation (GS31/32), and at the booting


217 stage (GS47–49), for a total seasonal nitrogen application of 180 kg N ha &#91;-&#93; ¹. Chemical weed


218 control was achieved using 5 g a.i. ha &#91;-&#93; ¹ florasulam + 45 g a.i. ha &#91;-&#93; ¹ pinoxaden (Axial


219 komplett®) and 20 g a.i. ha &#91;-&#93; ¹ carfentrazone-ethyl during post-emergence for grass weed


220 control, and eight days after using 700 g a.i. ha &#91;-&#93; ¹ mecoprop-P + 320 g a.i. ha &#91;-&#93; ¹ 2,4-D (Aurora


221 40 WG® and Duplosan KV neu®) for broadleaf weed control. Deltamethrin was applied at


222 7.5 g a.i. ha &#91;-&#93; ¹ (Decis Forte®) to control insect pests. The entire field trial was grown under


223 strictly rainfed conditions. Daily meteorological data, including ambient temperature,


8


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


224 precipitation, and relative humidity, were recorded by a dedicated weather station located 200


225 m from the plots (Figure S1).


226 2.3. Greenhouse experiment


227 A parallel experiment was conducted in the greenhouse facilities of IPK Gatersleben. Sowing


228 was performed 3 cm deep in individual cells of 24-well Quickpot trays (model 24T/16), with


229 one seed per cell, following a randomised complete block design (RCBD) across three


230 independent blocks (replications). Each growth cell measured 55 × 60 × 160 mm (volume


231 ~330 cm³). The growth substrate consisted of sand (particle size 0.20–0.63 mm), pre-mixed


232 with Osmocote® Pro 3-4M controlled-release fertiliser (19% N, 9% P � O �, 10% K � O + 2%


233 MgO + micronutrients; 3 to 4-month release profile) at a concentration of 8 g L &#91;-&#93; ¹ of sand.


234 Plants were maintained under a 16h light / 8h dark photoperiod. During the light period,


235 photosynthetic photon flux density of 900 µmol m &#91;-&#93; ² s &#91;-&#93; ¹ was provided at the canopy level. Air


236 temperature was regulated at 17 °C during the day and 15 °C at night. Root samples were


237 collected 8 weeks after sowing (960 GDD), once plants had reached the booting stage.


238 2.4. Root sampling and anatomical phenotyping


239 Root systems in both greenhouse and field environments were targeted for harvest at the


240 booting stage (GS45–49). To account for genotype-specific phenological variation, field


241 sampling dates were adjusted within a ±2-week window. Following the shovelomics


242 approach ( _49_, _50_ ), field root crowns were excavated from rows 2 and 5 of each 6-row plot to


243 ensure uniform stand density and avoid edge effects ( _51_ ). Excavated root crowns were soaked


244 in a dilute liquid-soap solution for 18 h to loosen soil aggregates, then gently rinsed with a


245 low-pressure water spray. Three representative plants were selected as biological replicates


246 (Figure 1). In parallel, greenhouse-grown plants were extracted from Quickpot trays and


247 washed sand off the root system with a gentle water spray.


248 The main culm root system was preserved in 70% ethanol at 4 °C until further sub-sampling


249 for the anatomical analysis. Main-culm nodal roots originating from the first four culm nodes


250 (N1 to N4, where N1 = oldest/most basal node, node 1, and N4 = youngest/most apical node,


251 node 4) were sampled sequentially. Transverse sections were cut precisely at a distance of


252 2 � cm from the root-shoot emergence junction ( _39_, _52_, _53_ ). The theoretical maximum


253 sampling size comprised 4,560 anatomical specimens (2 growth conditions × 190 genotypes


9


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


254 × 4 nodal positions × 3 biological replicates). We obtained 4242 anatomical specimens in


255 total for further image analysis (Table S2).


256 Sampled roots were mounted on the Rapid Anatomics Tool and imaged without chemical


257 staining using UV-induced cell-wall autofluorescence (excitation: 340-380 nm; emission:


258 &gt;420 nm). Digital RGB images were acquired in PNG format using a digital microscope


259 camera and Dino-Capture 2.0 (v1.553B) software ( _54_ ). Magnification was set to 190x-220x


260 for basal nodes (N1 and N2) and 175x-190x for apical nodes (N3 and N4) to account for


261 increasing root cross-sectional area (Figure 1).


262
263 Figure 1. Nodal root sampling strategy and anatomical phenotyping. (A) Schematic of the main culm


264 illustrating distinct nodal root classes (Nodes 1–4) and the sectioning zone, at 2 cm distal to the root

265 shoot junction. (B) Representative transverse root section highlighting the quantified anatomical


266 traits: root cross-sectional area, stele cross-sectional area, metaxylem vessel area and number, and


267 cortical cell file layer number.


268 Image processing and trait quantification were performed in Fiji (ImageJ v2.16.0) ( _55_ ).


269 Captured RGB images were converted to 8-bit, and ten spatial, meristic, and relational


270 anatomical traits were measured manually using region-of-interest (ROI) tools (Figure 1,


271 Table 1).


272 Table 1. Description of wheat root anatomical traits measured across nodal positions






|Traittype|Abbreviation|Units|Traitname|Derivation|
|---|---|---|---|---|
|Area|RxA|mm2|Root&lt;br&gt;cross-&lt;br&gt;sectional area|Measured|
|Area|SxA|mm2|Stele&lt;br&gt;cross-&lt;br&gt;sectional area|Measured|


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)







273 *Cortex width was calculated assuming circularity: �RxA/π ��SxA/π

|Area|CA|2&lt;br&gt;mm|Cortexarea|RxA SxA&lt;br&gt;-|
|---|---|---|---|---|
|Ratio|CSR|dimensionless|Cortex to stele&lt;br&gt;ratio|CA / SxA|
|Meristic|CCFN|count|Cortical cell file&lt;br&gt;layer number|Measured|
|Length|mCCD|mm|Mean&lt;br&gt;cortical&lt;br&gt;cell diameter|Cortex width*/ CCFN|
|Meristic|MXN|count|Metaxylem&lt;br&gt;vessel number|Measured|
|Area|tMXA|mm2|Total&lt;br&gt;metaxylem area|Measured|
|Area|mMXA|mm2|Mean&lt;br&gt;metaxylem area|tMXA / MXN|
|Ratio|MSR|dimensionless|Metaxylem&lt;br&gt;to&lt;br&gt;stele ratio|tMXA / SxA|
||||||



274 2.5. Allometric modelling and size-correction


275 2.5.1. Bivariate allometric modelling


276 To distinguish RxA-dependent from RxA-independent anatomical variation, bivariate scaling


277 relationships between individual traits ( � ) and root cross-sectional area (RxA, � ) were


278 modelled using Huxley’s power equation ( _56_ ):


��α �� &#91;�&#93; �� &#91;�&#93; 2. 1


279 � is the allometric coefficient, � is the scaling exponent, and e &#91;�&#93; represents multiplicative


280 error. For statistical estimation of scaling exponents, variables were linearised into base-10


281 logarithmic space:


��� �� ��������� �� ������� �� �α�� � 2. 2


282 We used a hybrid analytical protocol to address distinct biological and mathematical


283 objectives ( _44_ ). Standardised Major Axis (SMA) regression was used strictly to characterise


284 physiological scaling relationships because it accounts for biological variance in both


285 variables. However, because SMA slopes inflate when trait-size correlations are weak ( _b_ SMA


11


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


286 = _b_ OLS / ∣ r ∣ ), applying them directly to size-correction formulas yields unstable, over-corrected


287 phenotypes for low-R &#91;2&#93; traits (e.g., metaxylem-related traits, Table S4) ( _57_ ).


288 Therefore, the mathematical size-correction step utilised trait-specific statistical models at the


289 replicate level to accurately estimate the correction exponent ( � ). Continuous spatial traits


290 were modelled using ordinary least squares (OLS) regression in log 10 -log 10 space. Strictly


291 positive count variables (MXN, CCFN) were treated as discrete count traits rather than


292 categorical variables and modelled using Poisson or negative binomial generalised linear


293 models (GLMs) fitted directly to raw counts, with the resulting slope used as the allometric


294 scaling exponent. Although ratio traits (CSR, MSR) are already expressed relative to another


295 anatomical measurement, they may still vary with RxA. Therefore, ratio traits were not scaled


296 directly but were computed post-hoc from their independently size-corrected spatial


297 components.


298 2.5.2. Slope homogeneity


299 Before standardisation, an analysis of covariance (ANCOVA) was executed for each


300 environment-trait to formally test for slope homogeneity across nodal positions using the


301 model:


��� �� ������� �� ���������� 2. 3


302 A significant interaction term ( ��� �� ����� � ����, _P_ &lt; 0.05) confirmed heterogeneity in


303 scaling slopes across nodes (e.g., in a field dataset, CA scaling shifts from hyperallometry


304 &#91;b=1.52&#93; at Node 1 to isometry &#91;b=1.04&#93; at Node 4). Although some slope values were more


305 uniform in the greenhouse dataset, we utilised node-specific slope values to maintain


306 analytical consistency across both datasets (Figure 3, Table S3).


307 2.5.3. Allometric size correction


308 2.5.3.1. Node-specific allometric size correction for GWAS


309 To preserve the biological identity and ontogenetic context of each distinct nodal root


310 position, we used an environment- and node-specific common within-group standardisation


311 strategy. This node-specific reference baseline area was estimated using a geometric mean of


312 RxA ( � �,node (mm &#91;2&#93; ); N1 = 0.085, N2 = 0.144, N3 = 0.312, N4 = 0.394; for the field dataset;


313 Fig. S6) for roots from that specific nodal position and environment. Trait values were then


12


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


314 standardised (or size-corrected) to a node-specific baseline area using the allometric scaling


315 equation ( _44_ ) adapted to our case scenario:



� � �,node
adj,node ��� �



� node
2. 4



316 Transformed to a linear logarithmic format, the equation is expressed as:


��� �� !� ���,���� " ���� �� ����� ���� �#��� �� ������� �� !� �,���� "$ 2. 5


317 where � ���,���� is the size-corrected phenotype, � is the raw trait value, � is the raw


318 individual RxA at which the � trait value is measured, and � �,���� is the reference geometric


319 mean RxA for a given node. The scaling exponent, � ����, was fitted independently for each


320 combination of trait, node, and environment.


321 2.5.3.2. Common referenced size correction for


322 physiological/developmental comparisons


323 To facilitate the physiological comparison of anatomical trait trajectories (i.e., the sequential


324 patterns of change in a given trait across successive nodal cohorts from N1 to N4), a common


325 reference root cross-sectional area (RxA) within each environment is required. To avoid the


326 mathematical distortions of scaling roots outside their natural biological ranges, we identified


327 the range of overlapping RxA values among nodal positions separately for each environment


328 (0.09 to 0.39 mm &#91;2&#93; ; field) and set the common reference RxA ( � �,������ (mm &#91;2&#93; ); 0.198 for


329 field and 0.699 for greenhouse dataset; Figure S7) to the grand geometric mean RxA. Traits


330 were standardised using the node-specific scaling exponents ( � ���� ) projected to this shared


331 baseline within each environment.


��� �� !� adj,common " ���� �� ����� node �#��� �� ������� �� !� �,common "$ 2. 6


332 This allometric adjustment reduces the dependence of anatomical traits on RxA, and the


333 resulting traits are referred to as size-corrected traits for physiological comparisons. Thus,


334 node-specific reference RxA values were used for within-node genetic analyses, whereas a


335 common reference RxA within each environment was used to compare anatomical traits


336 across nodal positions.


13


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


337 2.5.3.3. Empirical validation and quality diagnostics for the size


338 correction protocol


339 To validate whether the standardisation pipeline has effectively removed the dependence of


340 stele and cortical traits on RxA in nodal roots, we did two trait-specific diagnostic checks.


341 Post-correction residual correlations ( _r_ ) between size-corrected traits and RxA were reduced


342 to near zero (| _r_ | ≤ 0.24) and were non-significant ( _P_ ≥ 0.05) for the vast majority of


343 combinations, confirming the removal of size-dependent confounding effect. Furthermore,


344 Spearman rank correlations ( ρ ) between raw and size-corrected entry means demonstrated


345 rank preservation ( ρ ≥ 0.75, _P &lt;_ 0.001) for metaxylem-related traits (mMXA, MSR, MXN,


346 tMXA) and mean cortical cell diameter (mCCD). In contrast, cortical area (CA) demonstrated


347 near-complete rank shuffling across all nodes ( ρ ranging from -0.30 to 0), mathematically


348 confirming that phenotypic variation in unadjusted CA is almost entirely driven by whole

349 root scaling rather than independent tissue partitioning (Figure S8).


350 2.6. Generalised Linear Mixed-Effects Models (GLMMs) for trait trends


351 Raw trait trajectories across nodes N1 to N4 were modelled using glmmTMB ( _58_ ) fitted on


352 replicate-level data:


� �,�,� ����� � �! 1 &amp; &#91;&amp; &#93; '�(�)*+� � " 2. 7


353 where � �,�,� is the phenotypic value of replicate ( of genotype � at nodal position, ( ���� � ;


354 fixed effect, N1–N4), and ! 1 &amp; &#91;&amp; &#93; '�(�)*+� � " represents the random intercept for genotype.


355 Error distributions were assigned by trait family: Gamma (log link) for the area and ratio


356 traits (RxA, CA, CSR, SxA, tMXA, mMXA, mCCD); Poisson (log link, auto-upgrading to


357 Negative Binomial if overdispersed) for count traits (MXN, CCFN); and Beta (logit link) for


358 bounded proportion traits (MSR). Pairwise node differences were evaluated via Tukey's HSD


359 test ( _P_ &lt; 0.05) within each trait.


360 2.7. Genotypic BLUE calculations, genotypic repeatability ( _R_ ) and GWAS


361 2.7.1. Environment-specific BLUE methodology


362 Genotypic Best Linear Unbiased Estimators (BLUEs) were calculated independently for


363 every environment-node-trait combination. For the field, as single-plot layouts confound plot


14


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


364 and genotype, BLUEs were calculated as the arithmetic mean of the three within-plot plant


365 subsamples per genotype:



-./0 �
�



�� &#91;∑&#93; ���� � �,� 2. 8



366 For the randomised complete block design (RCBD) in the greenhouse experiment, block

367 level variance was removed by fitting a fixed-effects linear model per node-trait combination:


� �,� �'�(�)*+� � ���+�,23)� � 2. 9


368 Where '�(�)*+� � is the fixed genotypic effect, and ��+�,23)� � is the fixed block effect.


369 Adjusted marginal means were extracted as BLUEs using emmeans (v2.0.4).


370 2.7.2. Genotypic repeatability ( � )


371 Environment-specific genotypic repeatability ( � ) was estimated independently for each trait

372 node combination using both raw unadjusted ( � ) and node-specific, size-corrected ( � ���,���� )


373 trait values. Variance components were extracted using random-effects models fitted with the


374 lme4 package ( _59_ ), matching the model structures used for the BLUE estimations described


375 above. Repeatability was calculated on an entry-mean basis, accounting for unbalanced data


376 ( _60_ ) as:



�
�� �

� ~~�~~ &#91;�&#93;&#91;�&#93;
� ����



2. 10



377 Where 4 � and 4 � are the genotypic and residual variances, respectively, and ( � is the


378 harmonic mean of replicates per genotype to adjust for unbalanced replication. Models that


379 produced a singular fit (where a variance component is estimated at the zero boundary) were


380 flagged but retained, as the resulting � estimates remain mathematically valid for quantitative


381 scaling (Figure S13).


382 We conducted two parallel GWAS analyses per trait, node (N1–N4) and environment


383 combination. GWAS using size-corrected traits was used to identify genetic associations with


384 anatomical variation after accounting for RxA, whereas GWAS using raw traits identified


385 associations with unadjusted anatomical variation, including variation associated with RxA.


15


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


386 2.7.3. Genotyping and Genome-Wide Association Studies (GWAS)


387 The Tango-Wheat panel was genotyped by TraitGenetics (SGS Institut Fresenius GmbH,


388 Gatersleben, Germany). For genomic DNA extraction, leaf tissue samples were taken from


389 young seedlings for each accession. The genotyping was performed using the 25K Infinium


390 iSelect array, providing 24,243 SNP markers after quality filtering (minor allele frequency


391 MAF &gt; 0.05, missingness &lt; 10%).


392 Genome-wide association studies were conducted in R using the Genome Association and


393 Prediction Integrated Tool (GAPIT v3.0; ( _61_ )) with both the Multiple Loci Mixed Model


394 (MLMM; ( _62_ )) and the Fixed and Random Model Circulating Probability Unification


395 (FarmCPU; ( _63_ )) models. MLMM was used for model validation.


396 To assess potential confounding factors, population structure and pairwise kinship were


397 characterised via SNP marker PCA (first four PCs) and the VanRaden matrix (VanRaden,


398 2008), respectively. To define marker-trait associations across all GWAS runs, a genome

399 wide threshold was established using LD-based marker pruning in PLINK v1.9 (r &#91;2&#93; = 0.2).


400 This retained M e = 4,402 independent SNPs out of 24,243 total markers, establishing a


401 suggestive genome-wide threshold of 1/M e ( _P_ = 2.27E-4, i.e., −log 10 _P_ ≥ 3.64) ( _64_ ).


402 2.8. Gene Ontology (GO) enrichment analysis and multi-omics candidate gene


403 prioritisation


404 To identify overrepresented genomic pathways among genes that met the suggestive


405 threshold, we performed Gene Ontology (GO) enrichment analysis. The gene stable IDs and


406 GO terms were retrieved from the Ensembl Plants BioMart database. To control for genomic


407 coverage biases inherent in SNP arrays and prevent false-positive enrichment, a local


408 background universe was established comprising all high-confidence wheat genes (IWGSC


409 RefSeq v2.1) located within a ±200 kb window of any tested SNP ( _65_ ). GO enrichment was


410 performed using a hypergeometric test via the enricher function in the R package


411 clusterProfiler ( _66_ ). Enriched terms were classified into "Plant vigour/growth" or "Tissue


412 allocation/patterning" via keyword matching (Table S11), and functional divergence between


413 the size-corrected and unadjusted raw models was evaluated using Chi-square tests.


414 To further identify candidate genes associated with loci detected in the GWAS, we evaluated


415 SNPs that exceeded the suggestive threshold (−log 10 _P_ ≥ 3.64) in the FarmCPU model. We


416 used an LD-aware windowing approach, in which genome-wide LD decay was estimated to


16


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


417 be ~5.3 Mb (r &#91;2&#93; = 0.2). Loci were initially defined using a core region of ±200 kb and an


418 extended window of ±2 Mb surrounding each lead SNP to retrieve primary encoded proteins


419 ( _65_ ). Candidate genes were subsequently prioritized through orthology mapping in


420 _Arabidopsis thaliana_ and rice ( _Oryza sativa_ ) using Ensembl Plants and Monocots PLAZA


421 5.0 ( _67_, _68_ ), functional annotations from UniProt (UniProt Consortium;


422 https://www.uniprot.org/), expression profiles from the Wheat eFP Browser ( _69_, _70_ ), and


423 documented functional associations with the target traits ( _71_ ).


424 To specifically investigate root-specific expression patterns of candidate genes, we utilised


425 the metadata table to annotate each run_accession with high-level plant organ designation:


426 “roots”, “leaves/shoot”, “spike”, or “grain”. Expression data were downloaded. For each gene


427 and organ category, the expression signal was calculated as the average TPM across the five


428 samples with the highest expression within that organ. Furthermore, genes with a maximum


429 organ-average TPM &lt; 10 were excluded to retain only moderately to highly root-expressed


430 candidates (requiring TPM ≥ 10) ( _72_ ).


431 2.9. Statistical analysis


432 All data processing, statistical modelling, and visualisation were performed in R (v.4.6.0)


433 using RStudio ( _73_ ). Data wrangling and ingestion were done with the tidyverse suite (v2.0.0),


434 while multivariate analysis &#91;principal component analysis (PCA)&#93; was done using


435 FactoMineR (v2.18) and factoextra (v2.2.0). Mixed-effects models were fitted using lme4


436 ( _59_ ) and glmmTMB ( _58_ ); with residual diagnostics performed using DHARMa (v0.5.0);


437 allometric scaling slopes were calculated using smatr ( _74_ ) (continuous traits) and MASS


438 (v7.3-66) (count traits, via Poisson/negative binomial regression); and genotypic BLUEs


439 were calculated using emmeans (v2.0.4). Genome-wide association analyses were performed


440 using GAPIT v3.0 ( _61_ ). Finally, data visualisations and post hoc mean comparisons (Tukey's


441 HSD) were generated using ggplot2 (v4.0.3), along with extension packages such as


442 patchwork (v1.3.2), ComplexUpset (v1.3.3), eulerr (v8.3.1), and multcompView (v0.1-12).


443 For downstream genomics, coordinate parsing and background universe construction were


444 handled via GenomicRanges (v1.64.0) and rtracklayer (v1.72.0), while GO enrichment and


445 comparative clustering were executed using clusterProfiler ( _66_ ).


17


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


446 **3.** **Results**


447 3.1. Unadjusted root anatomy varies systematically across successive nodal


448 positions


449 Before allometric correction for root cross-sectional area (RxA), root anatomical traits varied


450 systematically among roots emerging from the first four nodes of the main culm (N1–N4;


451 Figure 2). In the field, root cross-sectional area (RxA) increased from basal to apical nodal


452 positions (GLMM slope = 0.531, _P &lt;_ 0.001). Cortical area (CA), cortical cell file layer


453 number (CCFN), mean cortical cell diameter (mCCD), and cortex-to-stele area ratio (CSR)


454 also increased (slopes = 0.775, 0.400, 0.137, and 0.707, respectively; all _P &lt;_ 0.001). Stele


455 area (SxA) increased more modestly (slope = 0.039, _P &lt;_ 0.001). Metaxylem traits showed the


456 opposite trend across nodal root cohorts from N1 to N4. Total metaxylem area (tMXA), mean


457 metaxylem area (mMXA), and metaxylem-to-stele ratio (MSR) declined from N1 to N4


458 (slopes = −0.195, −0.210, and −0.261, respectively; all _P &lt;_ 0.001), whereas metaxylem


459 vessel number (MXN) did not change significantly (slope = 0.00864). Consequently,


460 increasing RxA across nodal positions was accompanied by greater cortical allocation and a


461 modest increase in stele area, whereas metaxylem area declined (Figure 2).


462 3.2. Root cross-sectional area is strongly associated with cortical traits


463 Across individual nodal positions, RxA was strongly correlated with CA (Pearson r = 0.95

464 1.00) and moderately to strongly correlated with SxA (r = 0.48 - 0.86; Figure S3).


465 Relationships between RxA and other cortical traits (CCFN, mCCD, and CSR) were more


466 variable, while metaxylem traits generally showed weaker associations (r = −0.32 to 0.67).


467 These results suggest that variation in RxA accounts for a substantial component of variation


468 in the cortical and stele dimensions, but is less consistently associated with metaxylem traits.


469 Principal component analysis (PCA) placed roots along a continuous gradient from basal


470 (N1) to more apical (N4) nodal positions (Figure 2). PC1 and PC2 explained 55.4% and


471 26.5% of variation, respectively. RxA and cortical traits contributed strongly to PC1, whereas


472 stele traits contributed predominantly to PC2. N1 and N4 occupied opposite ends of the main


473 gradient, with N2 and N3 in intermediate positions. Thus, nodal positions differed in their


474 overall anatomical phenotype, although the distributions overlapped rather than forming


475 discrete groups.


18


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


476 3.3. Nodal root anatomical trait trajectories are broadly maintained across growth


477 environments


478 Several positional trends were reproduced in greenhouse-grown plants (Figure S2). RxA, CA,


479 CCFN, SxA, and CSR generally increased from N1 to N4, whereas tMXA and MSR


480 declined. However, the magnitude of cortical changes was smaller in the greenhouse than in


481 the field. Metaxylem traits differed between environments. Mean metaxylem area declined in


482 both environments (field slope = −0.21; greenhouse slope = −0.087; both _P &lt;_ 0.001), but this


483 variation was associated with a decline in total metaxylem area in the field and an increase in


484 vessel number in the greenhouse. Greenhouse PCA similarly separated RxA and cortical


485 traits from vascular traits along the first two components (40.9% and 28.5% of variation,


486 respectively). These results indicate broad consistency in the direction of positional variation,


487 together with environmental differences in its magnitude and anatomical basis.


19


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


488
489 **Figure 2. Anatomical variation across successive nodal positions in field-grown wheat. Root**


490 **anatomical traits vary systematically among roots emerging from four successive nodes (N1–N4)**


491 **along the main culm, from basal to more apical positions.** (A) Principal component analysis (PCA)


492 of anatomical traits across four nodal positions. Genotype means were centred and estimated before


493 analysis. Points are coloured by nodal position, and ellipses represent 95% confidence regions.


494 Arrows indicate trait loading vectors, and axes show the percentage of variance explained by each


495 principal component. (B) Representative nodal root anatomical cross-sections at four nodal positions


496 (Node 1 - 4; basal to apical nodes). (C) Variation in root anatomical traits across nodal positions,


497 including root cross-sectional area (RxA), cortex area (CA), cortical cell file layer number (CCFN),


498 mean cortical cell diameter (mCCD), cortex-to-stele area ratio (CSR = CA/SxA), stele cross-sectional


499 area (SxA), metaxylem vessel number (MXN), total metaxylem area (tMXA), mean metaxylem area


500 (mMXA), and metaxylem-to-stele area ratio (MSR = tMXA/SxA). Boxplots show the distribution of


501 genotype means at each nodal position. Dashed red lines and shaded bands indicate predicted trends


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


502 and 95% confidence intervals from generalised linear mixed-effects models (GLMMs) fitted to


503 replicate-level data, with nodal position as a fixed effect and genotype as a random intercept. Gamma


504 models with a log link were used for continuous traits, Poisson models with a log link for count traits,


505 and a beta model with a logit link for MSR. Model slopes and associated P-values are shown in red.


506 Different lowercase letters indicate significant differences among nodal positions (Tukey's HSD, _P_ &lt;


507 0.05). The data represent 184 to 190 wheat genotypes, with each point corresponding to the mean of


508 three biological replicates. A detailed summary can be found in Table S5.


509 3.4. Anatomical traits show node-dependent allometric relationships


510 To determine how internal tissues scaled with RxA, we fitted allometric relationships


511 between RxA and individual anatomical traits at each nodal position (Figure 3). Cortex and


512 stele areas generally showed stronger RxA relationships than metaxylem traits. Fits for


513 CCFN and mCCD were intermediate (R² ≈ 0.18 to 0.64 for field and 0.11 to 0.46 for the


514 greenhouse dataset), whereas metaxylem traits (MXN, mMXA, and tMXA) frequently


515 showed weak or nonsignificant relationships, consistent with the greenhouse dataset (R² ≤


516 0.31). ANCOVA revealed significant interactions between RxA and nodal position,


517 indicating that the relationship between RxA and individual anatomical traits was not


518 constant across nodal positions on the main culm (Table S3, Figure S4). While specific


519 exponent values varied slightly between the field and greenhouse, their developmental trends


520 from N1 to N4 were consistent. Scaling exponents progressively decreased for cortical traits


521 (CA, mCCD, and CCFN) across successive nodes, but increased for metaxylem area traits


522 (tMXA and mMXA), with MXN remaining largely stable (Figure S5).


523


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


524 **Figure 3. Node-specific allometric scaling relationships for the field dataset.** Standardised Major


525 Axis (SMA) regressions of log10-transformed trait values against log10-transformed root cross

526 sectional area (RxA) across nodal positions (N1–N4). Annotations detail the SMA scaling exponent


527 ( _b_, i.e., _b_ node ) with 95% CI, variance explained (R²), and association significance (* _P_ &lt; 0.05, ** _P_ &lt;


528 0.01, *** _P_ &lt; 0.001, ns = not significant) for each node-trait combination. Trait abbreviations: CA,


529 cortex area; CCFN, cortical cell file layer number; mCCD, mean cortical cell diameter; SxA, stele


530 cross-sectional area; MXN, metaxylem vessel number; tMXA, total metaxylem area; mMXA, mean


531 metaxylem area.


532 3.5. Size correction reveals differences in tissue allocation among nodal positions


533 After standardisation to a common RxA ( � �,������ ), several anatomical trends persisted,


534 although their magnitude changed from the raw unadjusted trait values, where differences in


535 GLMM slopes were notably large (Figure 4, Figure S9, Figure S10). This allometric


536 correction reduced the dependence of anatomical traits on root cross-sectional area (RxA)


537 (Figure S8, Figure S11, Figure S12). The increase in cortical area was substantially


538 attenuated (Field: raw slope = 0.775; adjusted slope = 0.157; both _P &lt;_ 0.001). Similar


539 attenuation was observed for CSR (Field: 0.707 to 0.483), CCFN (Field: 0.4 to 0.217), and


540 MSR (Field: −0.261 to −0.184; all _P &lt;_ 0.001). Therefore, much of the raw variation in


541 cortical traits was associated with increasing RxA from basal to apical nodal positions.


542 Nevertheless, distinct positional differences remained at a common RxA.


543 The strongest change occurred in the stele area. SxA increased slightly across positions in the


544 raw data (slope = 0.039, _P &lt;_ 0.001), but declined after size correction (slope = −0.341, _P &lt;_


545 0.001) (Figure 4, Figure S9). This observation was also conserved in the greenhouse dataset


546 (slope = -0.014, _P &lt;_ 0.001) (Figure 4, Figure S10). At a common RxA, roots from more


547 apical nodes therefore had smaller steles than roots from basal nodes. Metaxylem traits


548 showed similar trends after correction for RxA. MXN was not observed to be associated with


549 nodal position in the raw data (Field: slope = 0.009, non-significant), but declined after


550 adjustment (slope = −0.136, _P &lt;_ 0.001). Consequently, the negative gradients in tMXA and


551 mMXA became much steeper after correction (−0.195 to −0.524 and −0.210 to −0.377,


552 respectively; all _P &lt;_ 0.001).


553 Furthermore, the size correction also distinguished two distinct cellular components of


554 cortical anatomy. While both CCFN and mCCD increased across nodal positions in the raw


555 data, size correction indicated that the increase in mCCD was largely associated with


556 variation in RxA; the size-corrected slope became non-significant in the field (slope = -0.007)


22


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


557 and significantly negative in the greenhouse (slope = -0.043, _P &lt;_ 0.001). Conversely, CCFN


558 continued to increase significantly across nodes in both environments (Field slope = 0.217;


559 Greenhouse slope = 0.050; both _P &lt;_ 0.001). These results suggest that greater cortical cell


560 file layer number, rather than increased mean cortical cell diameter, contributes more strongly


561 to positional differences in cortical organisation at a common RxA (Figure 4).


562 The size-corrected trait correlations revealed negative associations between several cortical


563 and stele traits across nodal positions (Figure S9, Figure S10). In the corresponding PCA


564 biplot of the field dataset, PC1 explained 72.1% of the variation and opposed cortical traits


565 (CA, CSR, CCFN) against vascular traits (tMXA, mMXA, SxA). PC2 explained 13.8% of


566 the variance and was associated mainly with MXN and mCCD. Nodal root cohorts (N1–N4)


567 remained largely separated along PC1, reflecting differences in cortical and vascular tissue


568 organisation across successive nodal positions. Together, these analyses support a continuous


569 difference in tissue allocation, with the clearest distinction between the most basal position


570 and more apical positions.


571 Together, these allometric adjustments reveal differences in tissue allocation across


572 successive nodal cohorts. At a common RxA ( ), roots emerging from more apical


573 nodes exhibited smaller stele and metaxylem areas and relatively greater cortical allocation.


574 The latter was associated primarily with increased cortical cell file layer number rather than


575 mean cortical cell diameter. The higher percentage of variance explained by PC1 in the field


576 (72.1%) compared with the greenhouse (47.7%) indicates that a greater proportion of


577 anatomical variation was captured by the first principal component under field conditions.


578
579 **Figure 4. Generalised linear mixed model (GLMM) slopes represent the trajectories of root**


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


580 **anatomical traits across nodes.** The plot contrasts the slopes of trait gradients across successive


581 developmental nodes (N1–N4) using raw (unadjusted) values versus size-corrected values in both


582 field and greenhouse environments, highlighting the directional shift in slopes after removing the


583 confounding effect of RxA. Trait abbreviations: CA, cortex area; CCFN, cortical cell file layer


584 number; mCCD, mean cortical cell diameter; CSR, cortex-to-stele area ratio (CA/SxA); SxA, stele


585 cross-sectional area; MXN, metaxylem vessel number; tMXA, total metaxylem area; mMXA, mean


586 metaxylem area; MSR, metaxylem-to-stele area ratio (tMXA/SxA). Significance levels for the


587 GLMM slopes: ns, not significant ( _P_ ≥ 0.05); _P_ &lt; 0.001.


588 3.6. Genetic associations differ between raw and size-corrected anatomical traits


589 Genotypic repeatability ( � ) for root anatomical traits varied significantly across


590 environments, nodes, and between raw and size-corrected measurements. Greenhouse


591 conditions yielded consistently higher _R_ estimates across most traits than the field


592 environment, where micro-environmental heterogeneity diluted the genetic signal,


593 particularly at early nodes (N1 and N2). Metaxylem and stele-related traits, specifically total


594 metaxylem area (tMXA), mean metaxylem area (mMXA), and the metaxylem-to-stele ratio


595 (MSR), exhibited the highest and most stable repeatability, peaking at more apical nodal


596 positions (N3 and N4) with greenhouse � values reaching 0.50 to 0.65. In contrast, cortical


597 cell file layer number (CCFN) and mean cortical cell diameter (mCCD) showed consistently


598 poor repeatability ( � &lt; 0.30) across all experiments. The application of size correction


599 generally caused a marginal reduction in � for structural traits relative to raw values,


600 confirming that a portion of the unadjusted genetic variance is driven by RxA. Nevertheless,


601 the size-corrected data retained robust genetic signals for key tissue-partitioning traits,


602 supporting their use in GWAS after accounting for variation in RxA (Figure S13).


603 Genetic variance estimates varied by trait, nodal position, and environment. Size correction


604 generally retained broad differences among traits in the reported repeatability estimates,


605 although some estimates changed considerably (Figure S13).


606 Genome-wide association analyses were conducted separately by trait, nodal position, and


607 environment (Figure S14, Table S7 and Table S8). Using the reported suggestive threshold


608 (−log 10 _P_ ≥ 3.64; FarmCPU), the raw-trait analyses identified 213 (164 unique between


609 environments) associated SNPs in the field and 132 (123 unique between environments) in


610 the greenhouse. In comparison, models utilising allometrically size-corrected traits detected


611 170 (162 unique) and 102 (79 unique) SNPs in the field and greenhouse, respectively. In the


612 raw field analysis, associations were distributed across cortical traits, vascular traits, and


24


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


613 RxA, with comparatively limited overlap among these categories. This limited intersection


614 between cortical and vascular trait associations persisted even after size correction.


615 The detected genetic associations for nodal root anatomy were highly trait- and environment

616 specific, with limited overlap among traits (Figure 5, Figure S15). Under field conditions,


617 associations were predominantly identified for RxA and cortical area traits, whereas


618 greenhouse associations were more frequently identified for metaxylem-related traits


619 (mMXA, tMXA), particularly at N4. Applying size corrections identified additional SNP


620 associations for nearly all measured traits that were not detected using raw anatomical


621 measurements. Although the sets of associated SNPs were largely distinct, some overlap


622 occurred between raw and size-corrected versions of the same traits, suggesting that some


623 associations were retained after accounting for variation in RxA. Few lead SNPs were shared


624 among nodal positions or between field and greenhouse environments. Thus, the detected


625 associations depended on trait, nodal position, and environment. However, limited overlap


626 among suggestive associations does not necessarily indicate distinct genetic control, as the


627 detection of associations also depends on allele effects, trait variation, linkage disequilibrium,


628 and statistical power.


25


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


629
630 **Figure 5. Shared and unique genetic loci associated with raw and size-corrected root anatomical**


631 **traits within the field environment.** UpSet plot illustrating the intersection of suggestive (−log10 P ≥


632 3.64; FarmCPU model) single-nucleotide polymorphisms (SNPs) across unadjusted (“_Raw”) and


633 size-corrected (“_SC”) traits. Vertical bars denote the number of SNPs unique to a specific trait

634 version combination or shared across multiple sets. Bar segments are coloured according to the


635 specific nodal positions (N1–N4) from which the significant associations were mapped. Horizontal


636 bars display the total number of suggestive SNPs detected for each individual trait version. Trait


637 abbreviations: CA, cortex area; CCFN, cortical cell file layer number; mCCD, mean cortical cell


638 diameter; CSR, cortex-to-stele area ratio (CA/SxA); SxA, stele cross-sectional area; MXN,


639 metaxylem vessel number; tMXA, total metaxylem area; mMXA, mean metaxylem area; MSR,


640 metaxylem-to-stele area ratio (tMXA/SxA).


641 3.7. Functional annotations differ between association sets


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


642 Candidate-gene sets assigned to raw and corrected associations partly overlapped, with 243


643 genes in the field and 169 in the greenhouse, as reported by the enrichment analysis (Figure


644 6). In the field, genes unique to the size-corrected analysis showed enrichment for functions


645 associated with cell-wall processes (e.g., lignin and cellulose biosynthesis) and transport


646 (e.g., carbohydrate and plasmodesmata-mediated transport). In contrast, genes from raw


647 associations more often included growth- and signalling-related terms (e.g., ethylene

648 activated signalling). A keyword-based comparison of these functional categories was


649 significant in the field (chi-square _P_ = 0.026), but not in the greenhouse (chi-square _P_ =


650 0.620) (Figure S16). These patterns suggest that trait adjustment changes the composition of


651 candidate-gene sets; functional classifications remain exploratory and do not establish causal


652 pathways.


653


654 **Figure 6. Functional enrichment and coherence of candidate genes under field conditions, where**


655 **all nodal cohorts and traits are consolidated.** (A) Side-by-side dot plot of significantly enriched


656 Biological Process (BP) Gene Ontology terms for the size-corrected (481 genes) and unadjusted raw


657 (457 genes) GWAS models. Node size represents the gene ratio, and colour indicates the Benjamini

658 Hochberg adjusted p-value. (B) Functional coherence of transformation-specific genes, illustrating the


659 total number of significant GO terms (Biological Process &#91;BP&#93;, Cellular Component &#91;CC&#93;, and


660 Molecular Function &#91;MF&#93;) derived from candidate genes uniquely identified by either the size

661 corrected or unadjusted models. The size-corrected unique gene set exhibits higher functional


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


662 coherence, yielding nearly twice as many significant BP terms as the unadjusted unique set. Details


663 about the greenhouse dataset are mentioned in Figure S16.


664 3.8 Representative genes from the raw and size-corrected data


665 Gene expression data revealed 76 root-expressed genes associated with allometrically size

666 adjusted traits and 64 root-expressed genes associated with raw, unadjusted root traits (Table


667 S10, Table S11), within the core SNP confidence interval of ±200 Kb across environments


668 and nodal root positions. Furthermore, we identified 32 genes shared between the allometric


669 and raw unadjusted trait sets, specifically associated with mCCD, mMXA, MSR, and MXN.


670 Interestingly, integrating GWAS associations with publicly available gene expression data


671 identified highly root-expressed candidate genes associated with anatomical traits across


672 different nodal positions, yielding partially distinct candidate gene sets from raw and size

673 corrected analyses.


674 Together, GWAS association studies suggest convergent and distinct regulation of RxA

675 independent and RxA-dependent root anatomical traits at different root node levels, a pattern


676 reported for the first time in cereals, specifically wheat.


677 **4.** **Discussion**


678 4.1. Successive nodal roots form a developmentally structured root system.


679 Cereal root systems are assembled progressively throughout plant development. Unlike


680 embryonic roots, which emerge during germination and the seedling stage, nodal roots


681 emerge sequentially from successive culm nodes as the plant develops ( _6_, _7_ ). This


682 organisation results in a mature root system composed of roots initiated at different stages of


683 whole-plant development. The contribution of nodal roots to the overall root system also


684 increases with plant age ( _75_, _6_, _76_, _17_ ). Thus, the progressive emergence of nodal roots


685 changes not only the composition of the root system but also the relative contribution of


686 different root classes throughout plant development.


687 Our results demonstrate that this developmental sequence is accompanied by systematic


688 changes in the anatomy of the roots produced. Roots emerging from successive main culm


689 nodes differed in RxA, cortical and stelar dimensions, and metaxylem anatomy. Root cross

690 sectional area and cortical allocation increased from basal to more apical nodal positions,


691 whereas total and mean metaxylem area declined. Multivariate analyses similarly revealed a


692 continuous gradient of anatomical variation across successive nodal positions rather than


28


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


693 discrete anatomical groups. These findings extend observations in maize, where roots


694 originating from successive stem nodes differ in anatomical and architectural traits ( _34_, _35_,


695 _37_ ), and indicate that anatomical differentiation within the post-embryonic root system is also


696 an important feature of wheat ( _8_, _25_, _77_ ).


697 An important distinction, however, is that nodal position is associated with several


698 developmental variables. Roots originating from basal nodes emerge earlier and are


699 consequently older at a common harvest stage, whereas roots from more apical nodes emerge


700 later and develop under a different whole-plant physiological context ( _6_, _78_ ). Nodal position,


701 root age, emergence time, and plant developmental stage are therefore not independent in the


702 present experiment. The observed anatomical gradients may consequently reflect differences


703 in root developmental age, the physiological conditions under which roots develop, or


704 position-associated developmental regulation and RxA ( _34_, _38_ ).


705 This organisation parallels the sequential development of aboveground organs. Cereal shoots


706 are composed of successive phytomers, with leaves and internodes at different positions


707 exhibiting distinct morphological and physiological characteristics ( _6_, _79_ – _81_ ). Extending this


708 developmental framework belowground suggests that nodal roots develop in structural


709 coordination with their corresponding parent phytomer ( _10_ ). Consistent with this


710 interpretation, we observed a positive relationship between stem diameter and root cross

711 sectional area (R² = 0.35, _P &lt;_ 0.001; Figure S17), indicating that the dimensions of nodal


712 roots are partly associated with those of the main culm node ( _6_, _82_ ). However, considerable


713 variation in RxA remained unexplained by stem diameter. Furthermore, roots from more


714 apical positions were larger than basal roots even within overlapping ranges of stem diameter.


715 These observations suggest that anatomical differences among nodal positions cannot be


716 attributed solely to differences in parent culm size. Because roots from successive nodal


717 positions emerge at different times, they also differ in age and developmental history.


718 Therefore, the anatomical differences observed here cannot be attributed exclusively to nodal


719 position, but may also reflect differences in root age, whole-plant developmental stage, and


720 the environmental conditions experienced during root development.


721 Together, these findings support a view of the wheat root system as a developmentally


722 organised series of anatomically distinct nodal roots. Rather than treating all post-embryonic


723 nodal roots as equivalent, their position within the developmental sequence should be


29


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


724 considered when characterising root phenotypes and interpreting their potential functions ( _34_,


725 _37_ ).


726 4.2. Allometric relationships reveal differences in tissue organisation among nodal


727 positions


728 A major challenge in interpreting anatomical variation among nodal roots is distinguishing


729 changes in tissue organisation from differences associated with RxA ( _44_, _83_ ). Larger roots


730 generally contain more cortical and vascular tissue, and the dimensions of individual tissues


731 often scale with root cross-sectional area ( _45_, _83_, _84_ ). Consequently, anatomical differences


732 among successive nodal positions could arise simply because later-emerging roots are larger.


733 This distinction is particularly important when comparing tissue dimensions across root


734 classes or developmental stages ( _34_, _85_ ).


735 Our allometric analysis showed that root cross-sectional area (RxA) accounted for a


736 substantial component of anatomical variation, particularly in cortical traits. The strong


737 increase in cortical area from N1 to N4 was considerably attenuated after size correction,


738 indicating that much of the observed cortical expansion was associated with increasing RxA.


739 However, several anatomical gradients persisted after standardisation to a common RxA. In


740 particular, the cortex-to-stele area ratio remained higher in roots from more apical nodes,


741 whereas stele area and metaxylem traits declined. These findings demonstrate that anatomical


742 variation across successive nodal positions cannot be explained by differences in RxA alone


743 (Figure 4, Figure S9, Figure S10).


744 The most pronounced change occurred in the stele area. Although the stele area increased


745 slightly from basal to apical nodes in the raw data, this relationship was reversed after size


746 correction (Figure 4). At a common RxA, roots originating from more apical nodes had


747 smaller steles than roots from basal nodes. Because total RxA constrains the space available


748 for individual tissues, greater cortical allocation necessarily corresponds to a smaller


749 proportion of the cross-section occupied by the stele. However, whether this reflects


750 differences in the developmental regulation of cortical and vascular tissues, or constraints on


751 their relative expansion, remains unclear. These results indicate that the relative organisation


752 of cortical and vascular tissues varies across successive nodal positions, even after accounting


753 for differences in RxA ( _83_, _84_ ).


30


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


754 Size correction also revealed differences in the cellular basis of cortical expansion. Both


755 cortical cell file number and mean cortical cell diameter increased across successive nodes in


756 the raw data. However, after accounting for RxA, the positional relationship with mean


757 cortical cell diameter was no longer significant, whereas cortical cell file number remained


758 positively associated with nodal position. Thus, at a comparable RxA, roots from more apical


759 nodes generally contained more cortical cell layers without a corresponding increase in mean


760 cortical cell diameter. This suggests that differences in cortical organisation are associated


761 more strongly with cortical cell file number than with radial enlargement of individual


762 cortical cells ( _34_, _37_, _86_ ). This distinction may have functional implications, as cortical cell


763 size and cell file number can influence the metabolic costs of root construction and


764 maintenance ( _23_, _30_ ). Larger cortical cells and fewer cortical cell files have been associated


765 with reduced metabolic costs, whereas additional cortical cell layers may contribute to root


766 mechanical properties and radial transport. Thus, the greater cortical cell file number


767 observed in roots from more apical nodes could reflect different functional requirements,


768 although these consequences were not directly evaluated in this study.


769 Importantly, the allometric relationships themselves also varied among nodal positions.


770 Cortical area showed a stronger increase with RxA in basal roots, whereas the scaling


771 exponents of several metaxylem traits were greater in roots from more apical nodes. This


772 indicates that the relationship between overall RxA and individual tissue dimensions is not


773 constant across the developmental sequence. Consequently, anatomical differentiation among


774 nodal positions involves both changes in absolute tissue dimensions and differences in how


775 tissues scale with RxA.


776 These patterns have potential functional implications. Cortical traits influence the metabolic


777 costs of root construction and maintenance, as well as radial transport and mechanical


778 properties, while stele and metaxylem dimensions are important determinants of hydraulic


779 transport ( _25_, _45_, _86_ ). The greater cortical allocation and reduced vascular dimensions in


780 roots emerging from more apical nodes at a common RxA could therefore influence the


781 balance between metabolic costs, mechanical properties, and water transport capacity. In


782 particular, differences in metaxylem vessel dimensions may influence axial hydraulic


783 conductance, which depends strongly on vessel radius. However, anatomical variation alone


784 does not establish functional differences. Future studies integrating anatomical allometry with


785 measurements of root respiration and theoretical axial hydraulic conductance, alongside


31


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


786 direct measurements of water and nutrient uptake, could help determine the functional


787 consequences of these anatomical differences.


788 4.3. Anatomical differentiation is broadly maintained across growth environments


789 The general trends in anatomical traits across successive nodal positions were broadly


790 consistent between field and greenhouse conditions. Root cross-sectional area, cortical area,


791 cortical cell file layer number, stele area, and cortex-to-stele ratio generally increased toward


792 more apical nodes, whereas the proportion of stele area occupied by metaxylem area


793 declined. Although the absolute trait values differed between environments, the broadly


794 similar trends from N1 to N4 suggest that positional variation is a recurring feature of wheat


795 nodal root development.


796 However, the magnitude of anatomical variation across nodal positions differed between


797 environments, particularly for cortical and metaxylem traits. Changes in cortical dimensions


798 across nodal positions were generally less pronounced in greenhouse-grown plants than in


799 field-grown plants (Figure 2, Figure S2). Although mean metaxylem area declined across


800 nodal positions in both environments, the associated changes in total metaxylem area and


801 vessel number differed. Total metaxylem area declined more strongly in the field, whereas


802 metaxylem vessel number remained relatively stable in the field but declined in the


803 greenhouse. Thus, similar positional trends in individual traits can be accompanied by


804 different patterns of anatomical variation between environments.


805 These differences may reflect the environmental conditions experienced during root


806 development. Successive nodal roots are produced at different times and may develop under


807 varying soil strength, temperature, water availability, and nutrient supply. Previous studies in


808 maize have similarly demonstrated that anatomical variation among nodal positions can


809 interact with environmental conditions, including nutrient and water availability ( _34_, _35_ ). Our


810 results suggest that anatomical trends across successive nodal positions are broadly consistent


811 between environments, although the magnitude of these trends and the underlying anatomical


812 characteristics can differ.


813 This variation also has implications for the interpretation of genotype-by-environment


814 interactions. Genetic differences in root anatomy may not be expressed uniformly across all


815 nodal positions. Instead, genotypes may differ in their overall anatomical phenotype, the


816 magnitude of positional variation, or the scaling relationships between individual tissues and


32


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


817 RxA ( _34_, _87_ ). Characterising these relationships across multiple nodal positions and


818 environments may therefore provide a more complete description of anatomical variation


819 than measurements from a single root position.


820 4.4. Integrating allometry into root phenotyping reveals distinct genetic


821 associations


822 From a phenotyping perspective, these results emphasise the importance of recording nodal


823 position and accounting for variation in RxA when comparing anatomical traits ( _37_, _88_ ) .


824 Measurements from a single nodal position may not represent the anatomical characteristics


825 of the mature root system, while pooling roots from different positions may obscure


826 systematic developmental variation ( _34_ ) . When comparing plant anatomical or architectural


827 traits, differences in overall size, such as length, area, volume, or biomass, can distort the data.


828 Allometric analysis allows anatomical traits to be compared after accounting for variation in


829 RxA. Integrating this approach with information on nodal position provides a framework for


830 identifying anatomical differences that are not explained by RxA alone. This may also help


831 distinguish genetic associations related to overall RxA from those associated with tissue


832 organisation.


833 Genetic effects on root dimensions can contribute indirectly to variation in cortical and


834 vascular traits ( _45_, _83_ ). Allometric correction provides a means of examining anatomical


835 variation after accounting for relationships with RxA ( _44_, _89_ ). Our parallel GWAS analyses


836 identified largely different associations for raw and size-corrected anatomical traits, with


837 limited overlap among nodal positions and between growth environments. This indicates that


838 the detected genetic associations depend on how anatomical phenotypes are defined and the


839 developmental position at which they are measured ( _34_, _85_ ). These differences demonstrate


840 that accounting for RxA can substantially alter which genomic regions are associated with


841 anatomical variation.


842 The functional annotations of candidate genes provided additional differences between the


843 two approaches. In field-grown plants, candidate genes identified exclusively by size

844 corrected analysis were enriched for functions associated with cell-wall processes and


845 transport, whereas raw trait associations more frequently included growth- and stress-related


846 functional categories. However, this distinction was less pronounced in greenhouse-grown


847 plants. These results suggest that size adjustment can influence the functional composition of


33


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


848 candidate-gene sets, although the biological significance of these differences remains


849 uncertain. Gene Ontology enrichment provides information about functional associations


850 among candidate genes but does not establish their causal involvement in anatomical


851 development ( _90_, _91_ ).


852 Among the identified loci, TRAESCS5A02G388300 was highly expressed in the roots and


853 associated with stele-related root traits, specifically in raw unadjusted MXN and


854 allometrically adjusted tMXA in nodes 3 and 2, respectively. This gene is orthologous to the


855 previously described AtBGLC1 in _Arabidopsis_ _thaliana_ _(AT5G20950)_, which is expressed in


856 both the primary and lateral root caps and regulates xyloglucan turnover, thereby promoting


857 primary cell-wall remodelling ( _92_, _93_ ) . TRAESCS6D02G406000 was identified as a highly


858 root-expressed, unique candidate associated with raw unadjusted mCCD in node 1.


859 TRAESCS6D02G406000 is described as CALS2/GSL3 in _Arabidopsis thaliana_


860 _(AT2G31960),_ which belongs to a 12-member family of CALLOSE SYNTHASE


861 (CALS)/GLUCAN SYNTHASE-LIKE (GSL) genes responsible for the biosynthesis of


862 callose ( β -1,3-glucan). Its paralog GSL8 was proposed to interact with PDLP5 to regulate


863 callose deposition at plasmodesmata ( _94_ ) . TRAESCS2B02G236500 was detected as highly


864 root-expressed and was a unique candidate associated with CCFN in node 3.


865 TRAESCS2B02G236500 ortholog gene in barley, HvCslf3 (HORVU2Hr1G042350), is part


866 of the Poaceae-specific CslF gene family responsible for the (1,4)- β -linked glucoxylan, which


867 forms the backbone of the hemicellulose in the primary cell wall. Interestingly, it was


868 reported that mutated CslF3 RNAi alleles in barley show reduced root length growth, a


869 significant reduction in root cortical area, and a trend toward reduced cortical cell file number


870 compared to the wild type ( _95_ ) .


871 The limited overlap among GWAS associations across nodal positions may also reflect


872 differences in the genetic regulation of root anatomy during plant development. However,


873 this interpretation requires caution. Differences in association detection can arise from


874 variation in trait distributions, genetic variance, environmental effects, and statistical power


875 ( _91_ ). Similarly, loci detected using size-corrected traits cannot automatically be interpreted as


876 controlling tissue patterning independently of overall root growth. Size correction removes


877 the modelled relationship with the root cross-sectional area but does not establish the


878 underlying genetic mechanism ( _44_, _89_ ). Functional validation and additional genetic analyses


879 will therefore be necessary to determine whether the associations identified here represent


880 distinct regulators of tissue organisation.


34


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


881 Together, these findings demonstrate the value of integrating nodal position and allometric


882 relationships into root anatomical phenotyping. Rather than treating nodal roots as equivalent


883 sampling units or relying exclusively on raw anatomical measurements, this approach allows


884 anatomical variation to be evaluated within its developmental context and relative to a


885 defined reference RxA. Combining raw and size-corrected traits with genetic analyses


886 provides a more comprehensive framework for investigating the genetic architecture of root


887 anatomy and identifying anatomical variation that may otherwise be obscured by differences


888 in overall root cross-sectional area.


889 **5.** **Conclusions**


890 Successive wheat nodal roots differ systematically in their anatomy, and these differences


891 cannot be explained by variation in RxA alone. Although increasing root cross-sectional area


892 accounts for much of the cortical expansion observed across successive nodal positions,


893 differences in cortical cell file number, stele area, and metaxylem anatomy persist after


894 allometric correction. Furthermore, the scaling relationships between RxA and individual


895 tissues vary among nodal positions, indicating that anatomical differentiation involves


896 changes in both tissue dimensions and their relative organisation.


897 These patterns are broadly maintained across greenhouse and field environments, although


898 their magnitude and anatomical basis differ. The distinct genetic associations identified for


899 raw and size-corrected traits further underscore the importance of accounting for allometry


900 when characterising the genetic architecture of root anatomy, as well as when investigating


901 the underlying developmental mechanisms and true physiological function of root traits.


902 Together, our findings support a view of the mature wheat root system as a developmentally


903 organised series of anatomically differentiated nodal roots rather than a collection of


904 equivalent root structures. Understanding how these anatomical differences influence


905 resource acquisition and root function will be important for identifying developmentally


906 relevant root traits for crop improvement.


907 **References**


908 1. P. R. Shewry, S. J. Hey, The contribution of wheat to human diet and health. _Food_


909 _and Energy Security_ 4, 178–202 (2015).


35


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


910 2. O. Erenstein, M. Jaleta, K. A. Mottaleb, K. Sonder, J. Donovan, H. Braun, M.


911 Reynolds, Wheat improvement: food security in a changing climate. _Global Trends in Wheat_


912 _Production, Consumption and Trade_, 47–66 (2022).


913 3. D. H. Jones, A. B. Tawale, H. M. Schneider, An introduction to the fundamentals of


914 root systems research in cultivated plants. _Ann Bot_ 138, 291–306 (2026).


915 4. R. Nirmalaruban, R. Yadav, S. S., A. Meda, P. Babu, M. Kumar, K. B. Gaikwad, N.


916 K. Bainsla, S. K. Singh, S. R., M. Rahimi, Root traits: a key for breeding climate-smart wheat


917 (triticum aestivum). _Plant Breeding_ 144, 310–334 (2025).


918 5. P. W. Barlow, “The origin, diversity and biology of shoot-borne roots” in _Biology of_


919 _Adventitious Root Formation_, T. D. Davis, B. E. Haissig, Eds. (Springer US, Boston, MA,


920 1994; https://doi.org/10.1007/978-1-4757-9492-2_1), pp. 1–23.


921 6. B. Klepper, R. Belford, R. Rickman, Root and shoot development in winter Wheat1.


922 _Agronomy Journal_ 76, 117–122 (1984).


923 7. F. Hochholdinger, K. Woll, M. Sauer, D. Dembinsky, Genetic dissection of root


924 formation in maize (zea mays) reveals root � type specific developmental programmes. _Ann_


925 _Bot_ 93, 359–368 (2004).


926 8. C. Bellini, D. I. Pacurar, I. Perrone, Adventitious roots and lateral roots: similarities


927 and differences. _Annual Review of Plant Biology_ 65, 639–666 (2014).


928 9. E. J. M. Kirby, M. Appleyard, N. A. Simpson, Co-ordination of stem elongation and


929 Zadoks growth stages with leaf emergence in wheat and barley. _The Journal of Agricultural_


930 _Science_ 122, 21–29 (1994).


931 10. M. Schierenbeck, A. B. Tawale, I. Lopez-Valdivia, D. H. Jones, A. Wolf, P. Linow,


932 C. Trautewig, H. M. Schneider, A thermal time framework drives coordinated below- and


933 above-ground development in temperate cereal crops. bioRxiv &#91;Preprint&#93; (2026).


934 https://doi.org/10.64898/2026.04.24.720680.


935 11. A. Fusseder, The longevity and activity of the primary root of maize. _Plant Soil_ 101,


936 257–265 (1987).


36


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


937 12. T. Knipfer, W. Fricke, Water uptake by seminal and adventitious roots in relation to


938 whole-plant water flow in barley (Hordeum vulgare L.). _J Exp Bot_ 62, 717–733 (2011).


939 13. C. E. Tunc, N. von Wirén, Hidden aging: the secret role of root senescence. _Trends in_


940 _Plant Science_ 30, 553–564 (2025).


941 14. G. O. Boatwright, H. Ferguson, Influence of primary and/or adventitious root systems


942 on wheat production and nutrient uptake. _Agronomy Journal_ 59, 299–302 (1967).


943 15. M. A. Ahmed, M. Zarebanadkouki, F. Meunier, M. Javaux, A. Kaestner, A.


944 Carminati, Root type matters: measurement of water uptake by seminal, crown, and lateral


945 roots in maize. _J Exp Bot_ 69, 1199–1206 (2018).


946 16. M. Schierenbeck, M. C. Fleitas, F. Cortese, S. I. Golik, M. R. Simón, Nitrogen


947 accumulation in grains, remobilization and post-anthesis uptake under tan spot and leaf rust


948 infections on wheat. _Field Crops Research_ 235, 27–37 (2019).


949 17. H. M. Schneider, J. A. Postma, J. Kochs, D. Pflugfelder, J. P. Lynch, D. van


950 Dusschoten, Spatio-temporal variation in water uptake in seminal and nodal root systems of


951 barley plants grown in soil. _Front. Plant Sci._ 11 (2020).


952 18. Y. Jiang, J. K. Whalen, Crown roots dominate nitrate uptake in maize (Zea mays)


953 under varying soil moisture conditions. _Ann Bot_ 138, 1889–1900 (2026).


954 19. I. Krassovsky, Physiological activity of the seminal and nodal roots of crop plants.


955 _Soil Science_ 21, 307 (1926).


956 20. T. D. Colmer, H. Greenway, Ion transport in seminal and adventitious roots of cereals


957 during O2 deficiency. _J Exp Bot_ 62, 39–57 (2011).


958 21. Z. Liu, R. Giehl, A. Hartmann, M. Hajirezaei, S. Carpentier, N. Von Wirén, Seminal


959 and nodal roots of barley differ in anatomy, proteome and nitrate uptake capacity. _Plant &amp;_


960 _cell physiology_, doi: 10.1093/pcp/pcaa059 (2020).


961 22. J. P. Lynch, Root phenes that reduce the metabolic costs of soil exploration:


962 opportunities for 21st century agriculture. _Plant, Cell &amp; Environment_ 38, 1775–1784 (2015).


37


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


963 23. I. Lopez-Valdivia, X. Yang, J. P. Lynch, Large root cortical cells and reduced cortical


964 cell files improve growth under suboptimal nitrogen in silico. _Plant Physiol_ 192, 2261–2275


965 (2023).


966 24. G. M. Whiteley, J. S. Hewitt, A. R. Dexter, The buckling of plant roots. _Physiologia_


967 _Plantarum_ 54, 333–342 (1982).


968 25. T. Colombi, A. M. Herrmann, P. Vallenback, T. Keller, Cortical cell diameter is key


969 to energy costs of root growth in wheat. _Plant Physiol_ 180, 2049–2060 (2019).


970 26. D. J. Vanhees, K. W. Loades, A. G. Bengough, S. J. Mooney, J. P. Lynch, Root


971 anatomical traits contribute to deeper rooting of maize under compacted field conditions.


972 _Journal of Experimental Botany_ 71, 4243–4257 (2020).


973 27. B. Hu, A. Henry, K. M. Brown, J. P. Lynch, Root cortical aerenchyma inhibits radial


974 nutrient transport in maize (Zea mays). _Ann Bot_ 113, 181–189 (2014).


975 28. H. M. Schneider, T. Wojciechowski, J. A. Postma, K. M. Brown, A. Lücke, V.


976 Zeisler, L. Schreiber, J. P. Lynch, Root cortical senescence decreases root respiration,


977 nutrient content and radial water and nutrient transport in barley. _Plant, Cell &amp; Environment_


978 40, 1392–1408 (2017).


979 29. R. Richards, J. Passioura, A breeding program to reduce the diameter of the major


980 xylem vessel in the seminal roots of wheat and its effect on grain yield in rain-fed


981 environments. _Australian Journal of Agricultural Research_ 40, 943–950 (1989).


982 30. J. P. Lynch, C. F. Strock, H. M. Schneider, J. S. Sidhu, I. Ajmera, T. Galindo

983 Castañeda, S. P. Klein, M. T. Hanlon, Root anatomy and soil resource capture. _Plant and Soil_


984 466, 21–63 (2021).


985 31. L. Barry, J. C. B. Cabrera, M. Lucas, G. Lobet, Y. Boursiac, A. Grondin, Role of


986 xylem in root hydraulics: Functionality and implications for drought adaptation. _Quantitative_


987 _Plant Biology_ 6, e42 (2025).


988 32. H. Tai, X. Lu, N. Opitz, C. Marcon, A. Paschold, A. Lithio, D. Nettleton, F.


989 Hochholdinger, Transcriptomic and anatomical complexity of primary, seminal, and crown


38


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


990 roots highlight root type-specific functional diversity in maize (Zea mays L.). _J Exp Bot_ 67,


991 1123–1135 (2016).


992 33. C. Clément, H. M. Schneider, D. B. Dresbøll, J. P. Lynch, K. Thorup-Kristensen,


993 Root and xylem anatomy varies with root length, root order, soil depth and environment in


994 intermediate wheatgrass (Kernza®) and alfalfa. _Ann Bot_ 130, 367–382 (2022).


995 34. J. T. Yang, H. M. Schneider, K. M. Brown, J. P. Lynch, Genotypic variation and


996 nitrogen stress effects on root anatomy in maize are node specific. _Journal of Experimental_


997 _Botany_ 70, 5311 (2019).


998 35. T. Koehler, Y. Kim, S.-Y. Tung, A. Heymans, N. Tyborski, F. Steiner, A. J. Wild, J.


999 Pausch, M. A. Ahmed, H. M. Schneider, Node of origin matters: comparative analysis of soil


1000 water limitation effects on nodal root anatomy in maize (Zea mays). _Ann Bot_ 136, 1031–1046


1001 (2025).


1002 36. E. Kirby, Botany of the wheat plant. _Bread Wheat. Improvement and Production._


1003 _Food and Agriculture Organization of the United Nation. Rome_, 19–37 (2002).


1004 37. L. York, J. Lynch, Intensive field phenotyping of maize (Zea mays L.) root crowns


1005 identifies phenes and phene integration associated with plant growth and nitrogen acquisition.


1006 _Journal of Experimental Botany_ 66, 5493–5505 (2015).


1007 38. R. S. Poethig, Vegetative phase change and shoot maturation in plants. _Current topics_


1008 _in developmental biology_ 105, 125–152 (2013).


1009 39. Md. N. Siddiqui, A. M. Ambaw, A. Shrestha, K. Pandey, T. J. Teferi, M. T. Gabi, D.


1010 Duarte-Delgado, J. Léon, A. Ballvora, Natural genetic variation in nodal root growth angle


1011 and anatomy underlies drought tolerance in bread wheat. _Environmental and Experimental_


1012 _Botany_ 237, 106220 (2025).


1013 40. D. S. Glazier, Complications with body-size correction in comparative biology:


1014 possible solutions and an appeal for new approaches. _J Exp Biol_ 225, jeb243313 (2022).


1015 41. D. W. Thompson, On growth and form. _Cambridge University_ (1936).


1016 42. S. J. Gould, Allometry and size in ontogeny and phylogeny. _Biological Reviews_ 41,


1017 587–638 (1966).


39


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


1018 43. J. D. Reist, An empirical evaluation of several univariate methods that adjust for size


1019 variation in morphometric data. _Can. J. Zool._ 63, 1429–1439 (1985).


1020 44. J. Lleonart, J. Salat, G. J. Torres, Removing allometric effects of body size in


1021 morphological analysis. _Journal of Theoretical Biology_ 205, 85–93 (2000).


1022 45. A. L. Burton, K. M. Brown, J. P. Lynch, Phenotypic diversity of root anatomical and


1023 architectural traits in zea species. _Crop Science_ 53, 1042–1055 (2013).


1024 46. Y. Zhang, J.-J. Cao, Q.-P. Yang, M.-Z. Wu, Y. Zhao, D.-L. Kong, The worldwide


1025 allometric relationship in anatomical structures for plant roots. _Plant Diversity_ 45, 621–629


1026 (2023).


1027 47. M. Altermann, J. Rinklebe, I. Merbach, M. Körschens, U. Langer, B. Hofmann,


1028 Chernozem—Soil of the Year 2005. _Journal of Plant Nutrition and Soil Science_ 168, 725–


1029 740 (2005).


1030 48. J. C. Zadoks, T. T. Chang, C. F. Konzak, A decimal code for the growth stages of


1031 cereals. _Weed Research_ 14, 415–421 (1974).


1032 49. S. Trachsel, S. M. Kaeppler, K. M. Brown, J. P. Lynch, Shovelomics: high throughput


1033 phenotyping of maize (Zea mays L.) root architecture in the field. _Plant Soil_ 341, 75–87


1034 (2011).


1035 50. L. M. York, S. Slack, M. J. Bennett, M. J. Foulkes, Wheat shovelomics I: A field


1036 phenotyping approach for characterising the structure and function of root systems in tillering


1037 species. bioRxiv &#91;Preprint&#93; (2018). https://doi.org/10.1101/280875.


1038 51. T. Colombi, N. Kirchgessner, C. A. Le Marié, L. M. York, J. P. Lynch, A. Hund,


1039 Next generation shovelomics: set up a tent and REST. _Plant Soil_ 388, 1–20 (2015).


1040 52. J. Liu, M. Dai, S. Liu, Yue, Z. Qin, C. Liu, R. Wang, Anatomical and Physiological


1041 Responses of Maize Nodal Roots to Shading Stress and Nitrogen Supply. _Agronomy_, doi:


1042 10.3390/agronomy15081949 (2025).


1043 53. J. C. Baca Cabrera, D. H. Jones, J. Vanderborght, D. Behrend, H. M. Schneider, G.


1044 Lobet, Root anatomical gradients and cultivar differences underlie variation in root hydraulic


1045 properties in German winter wheat. _J Exp Bot_, erag177 (2026).


40


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


1046 54. D. H. Jones, J. C. Baca Cabrera, D. Behrend, D. M. Wells, J. F. Swift, J. A. Atkinson,


1047 M. Schön, G. Lobet, M. T. Hanlon, H. M. Schneider, The Rapid Anatomics Tool (RAT): A


1048 low-cost root anatomical phenotyping platform reveals changes in root anatomy along the


1049 root axis. _Plant Phenomics_ 8, 100150 (2026).


1050 55. J. Schindelin, I. Arganda-Carreras, E. Frise, V. Kaynig, M. Longair, T. Pietzsch, S.


1051 Preibisch, C. Rueden, S. Saalfeld, B. Schmid, J.-Y. Tinevez, D. J. White, V. Hartenstein, K.


1052 Eliceiri, P. Tomancak, A. Cardona, Fiji: an open-source platform for biological-image


1053 analysis. _Nat Methods_ 9, 676–682 (2012).


1054 56. J. Huxley, Problems of relative growth Methuen &amp; co. _Ltd., London_ 104 (1932).


1055 57. D. I. Warton, I. J. Wright, D. S. Falster, M. Westoby, Bivariate line-fitting methods


1056 for allometry. _Biological reviews_ 81, 259–291 (2006).


1057 58. M. McGillycuddy, G. Popovic, B. M. Bolker, D. I. Warton, Parsimoniously fitting


1058 large multivariate random effects in glmmTMB. _J. Stat. Soft._ 112, 1–19 (2025).


1059 59. D. Bates, M. Mächler, B. Bolker, S. Walker, Fitting linear mixed-effects models using


1060 lme4. _Journal of Statistical Software_ 67, 1–48 (2015).


1061 60. J. Holland, W. E. Nyquist, C. T. Cervantes-Martinez, Estimating and interpreting


1062 heritability for plant breeding: Anupdate. _Plant breeding reviews_ 22, 9–111 (2003).


1063 61. J. Wang, Z. Zhang, GAPIT Version 3: Boosting Power and Accuracy for Genomic


1064 Association and Prediction. _genom. proteom. bioinform._ 19, 629–640 (2021).


1065 62. V. Segura, B. J. Vilhjálmsson, A. Platt, A. Korte, Ü. Seren, Q. Long, M. Nordborg,


1066 An efficient multi-locus mixed model approach for genome-wide association studies in


1067 structured populations. _Nat Genet_ 44, 825–830 (2012).


1068 63. X. Liu, M. Huang, B. Fan, E. S. Buckler, Z. Zhang, Iterative usage of fixed and


1069 random effect models for powerful and efficient genome-wide association studies. _PLOS_


1070 _Genetics_ 12, e1005767 (2016).


1071 64. P. Duggal, E. M. Gillanders, T. N. Holmes, J. E. Bailey-Wilson, Establishing an


1072 adjusted p-value threshold to control the family-wide type 1 error in genome wide association


1073 studies. _BMC Genomics_ 9, 516 (2008).


41


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


1074 65. T. Zhu, L. Wang, H. Rimbert, J. C. Rodriguez, K. R. Deal, R. De Oliveira, F. Choulet,


1075 G. Keeble-Gagnère, J. Tibbits, J. Rogers, K. Eversole, R. Appels, Y. Q. Gu, M. Mascher, J.


1076 Dvorak, M. C. Luo, Optical maps refine the bread wheat Triticum aestivum cv. Chinese


1077 Spring genome assembly. _The Plant Journal_ 107, 303–314 (2021).


1078 66. T. Wu, E. Hu, S. Xu, M. Chen, P. Guo, Z. Dai, T. Feng, L. Zhou, W. Tang, L. Zhan,


1079 X. Fu, S. Liu, X. Bo, G. Yu, clusterProfiler 4.0: A universal enrichment tool for interpreting


1080 omics data. _Innovation (Camb)_ 2, 100141 (2021).


1081 67. M. Van Bel, F. Silvestri, E. M. Weitz, L. Kreft, A. Botzki, F. Coppens, K.


1082 Vandepoele, PLAZA 5.0: extending the scope and power of comparative and functional


1083 genomics in plants. _Nucleic Acids Res_ 50, D1468–D1474 (2022).


1084 68. A. D. Yates, O. Austine-Orimoloye, A. G. Azov, M. Barba, I. Barnes, V. P. Barrera

1085 Enriquez, A. Becker, R. Bennett, A. Berry, J. Bhai, S. K. Bhurji, P. R. Branco Lins, L.


1086 Brooks, S. Budhanuru Ramaraju, L. I. Campbell, M. Carbajo Martinez, J. Carpenter, M.


1087 Charkhchi, L. A. Cortes, C. Davidson, S. Dickson, K. Dodiya, S. Donaldson, B.


1088 El Houdaigui, T. El Naboulsi, A. Fairbrother-Browne, O. Falola, R. Fatima, J.


1089 Gonzalez Martinez, T. Gurbich, H. Hall, M. Hardy, Z. Hollis, T. Hunt, M. Kay, V. Kaikala,


1090 A. Lazar, D. Lemos, D. Lodha, N. Mathlouthi, G. A. Merino, R. Merritt, L. P. Mirabueno, A.


1091 Mushtaq, S. Nakib Hossain, J. G. Pérez-Silva, I. Piližota, D. Poppleton, I. Prosovetskaia, S.


1092 Raj, A. I. A. Salam, S. Saraf, S. Sinha, B. Sipos, V. Sitnik, M.-M. Suner, L. Surapaneni, J. A.


1093 S. Tierney, D. Urbina-Gómez, A. Veidenberg, T. A. Walsh, J. M. Allen, J. Alvarez-Jarreta, J.


1094 Cheema, J. Batista da Rocha, N. H. De Silva, F. F. Tricomi, S. Giorgetti, G. R. Ilsley, J.


1095 Keatley, J. E. Loveland, J. M. Mudge, G. Naamati, J. Tate, N. L. Willhoft, A. Winterbottom,


1096 B. R. Flint, A. Frankish, L. Haggerty, S. E. Hunt, E. L. Clark, S. C. Dyer, M. A. Freeberg, F.


1097 J. Martin, R. D. Finn, Ensembl 2026. _Nucleic Acids Res_ 54, D1053–D1060 (2026).


1098 69. P. Borrill, R. Ramirez-Gonzalez, C. Uauy, expVIP: a Customizable RNA-seq Data


1099 Analysis and Visualization Platform1&#91;OPEN&#93;. _Plant Physiol_ 170, 2172–2186 (2016).


1100 70. R. H. Ramírez-González, P. Borrill, D. Lang, S. A. Harrington, J. Brinton, L.


1101 Venturini, M. Davey, J. Jacobs, F. van Ex, A. Pasha, Y. Khedikar, S. J. Robinson, A. T. Cory,


1102 T. Florio, L. Concia, C. Juery, H. Schoonbeek, B. Steuernagel, D. Xiang, C. J. Ridout, B.


1103 Chalhoub, K. F. X. Mayer, M. Benhamed, D. Latrasse, A. Bendahmane, International Wheat


1104 Genome Sequencing Consortium, B. B. H. Wulff, R. Appels, V. Tiwari, R. Datla, F. Choulet,


42


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


1105 C. J. Pozniak, N. J. Provart, A. G. Sharpe, E. Paux, M. Spannagl, A. Bräutigam, C. Uauy, The


1106 transcriptional landscape of polyploid wheat. _Science_ 361, eaar6089 (2018).


1107 71. Q. Yao, B. Chen, M. N. Lohani, Q. Chen, T. Wang, J. Chen, C. Wang, H. Zhang, L.


1108 Lu, A. Habib, J. Ma, Q. Xie, Genetic identification and characterization of putatively novel,


1109 stable, and validated loci for plant height in wheat (Triticum aestivum L.). _BMC Genomics_


1110 27, 347 (2026).


1111 72. X. Wang, P. Zhao, X. Shi, X. Guo, Y. Liu, W. Hou, M. Cheng, X. Liu, X. Lai, J.


1112 Simmonds, W. Harwood, J. Wang, Z. Liu, L. Huang, D. Han, W. Ji, C. Uauy, J. Xiao, Z.


1113 Kang, S. Xu, Population transcriptome and phenotype analyses reveal that Rht-D1b


1114 contributed a larger seedling root to modern bread wheat. _Plant Cell_ 37, koaf267 (2025).


1115 73. R Core Team, _R: A Language and Environment for Statistical Computing_ (R


1116 Foundation for Statistical Computing, Vienna, Austria, 2026; https://www.R-project.org/).


1117 74. D. I. Warton, R. A. Duursma, D. S. Falster, S. Taskinen, smatr 3– an R package for


1118 estimation and inference about allometric lines. _Methods in Ecology and Evolution_ 3, 257–


1119 259 (2012).


1120 75. P. Gregory, M. McGowan, P. Biscoe, Water relations of winter wheat: 2. Soil water


1121 relations. _The Journal of Agricultural Science_ 91, 103–116 (1978).


1122 76. J. Sebastian, M.-C. Yee, W. Goudinho Viana, R. Rellán-Álvarez, M. Feldman, H. D.


1123 Priest, C. Trontin, T. Lee, H. Jiang, I. Baxter, Grasses suppress shoot-borne roots to conserve


1124 water during drought. _Proceedings of the National Academy of Sciences_ 113, 8861–8866


1125 (2016).


1126 77. L. T. Abushal, I. W. McCahill, E. Z. Li, L. Pollard, C. F. Probert, S. P. Hazen, A


1127 hydroponic cultivation platform for analyzing nodal root development, cell wall thickening,


1128 and hormone responses in grasses. _Plant Methods_, doi: 10.1186/s13007-026-01586-7 (2026).


1129 78. M. E. McCully, Roots in soil: unearthing the complexities of roots and their


1130 rhizospheres. _Annual review of plant biology_ 50, 695–718 (1999).


1131 79. G. S. McMaster, Phytomers, phyllochrons, phenology and temperate cereal


1132 development. _The Journal of Agricultural Science_ 143, 137–150 (2005).


43


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


1133 80. Y. Huang, A. Maurer, R. F. Giehl, S. Zhao, G. Golan, V. Thirulogachandar, G. Li, Y.


1134 Zhao, C. Trautewig, A. Himmelbach, Dynamic phytomeric growth contributes to local


1135 adaptation in barley. _Molecular Biology and Evolution_ 41, msae011 (2024).


1136 81. T. Dresselhaus, M. Balboni, L. Berg, A. Dolata, F. Hochholdinger, Y. Huang, G. -q.


1137 Jiang, M. Von Korff, J.-C. Ku, K. Van Der Linde, J. Maika, C. L. Mondragon, M. Raissig, A.


1138 Schnittger, T. Schnurbusch, R. Simon, Y. Stahl, M. Timmermans, V. Thirulogachandar, S.


1139 Zhao, Y. Zhou, How meristems shape plant architecture in cereals—Cereal Stem Cell


1140 Systems (CSCS) Consortium. _The Plant Cell_ 37 (2025).


1141 82. K. J. Niklas, _Plant Allometry: The Scaling of Form and Process_ (University of


1142 Chicago Press, 1994).


1143 83. S. Wahl, P. Ryser, Root tissue structure is linked to ecological strategies of grasses.


1144 _New Phytologist_ 148, 459–471 (2000).


1145 84. D. Kong, C. Ma, Q. Zhang, L. Li, X. Chen, H. Zeng, D. Guo, Leading dimensions in


1146 absorptive root trait variation across 96 subtropical forest species. _New Phytologist_ 203, 863–


1147 872 (2014).


1148 85. S. P. Klein, H. M. Schneider, A. C. Perkins, K. M. Brown, J. P. Lynch, Multiple


1149 integrated root phenotypes are associated with improved drought tolerance. _Plant Physiol_


1150 183, 1011–1025 (2020).


1151 86. J. G. Chimungu, K. M. Brown, J. P. Lynch, Large root cortical cell size improves


1152 drought tolerance in maize. _Plant Physiol_ 166, 2166–2178 (2014).


1153 87. H. M. Schneider, V. S. N. Lor, M. T. Hanlon, A. Perkins, S. M. Kaeppler, A. N.


1154 Borkar, R. Bhosale, X. Zhang, J. Rodriguez, A. Bucksch, M. J. Bennett, K. M. Brown, J. P.


1155 Lynch, Root angle in maize influences nitrogen capture and is regulated by calcineurin B-like


1156 protein (CBL)-interacting serine/threonine-protein kinase 15 (ZmCIPK15). _Plant, Cell &amp;_


1157 _Environment_ 45, 837–853 (2022).


1158 88. G. T. Freschet, L. Pagès, C. M. Iversen, L. H. Comas, B. Rewald, C. Roumet, J.


1159 Klimešová, M. Zadworny, H. Poorter, J. A. Postma, T. S. Adams, A. Bagniewska-Zadworna,


1160 A. G. Bengough, E. B. Blancaflor, I. Brunner, J. H. C. Cornelissen, E. Garnier, A. Gessler, S.


1161 E. Hobbie, I. C. Meier, L. Mommer, C. Picon-Cochard, L. Rose, P. Ryser, M. Scherer

44


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766962; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766962)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


1162 Lorenzen, N. A. Soudzilovskaia, A. Stokes, T. Sun, O. J. Valverde-Barrantes, M. Weemstra,


1163 A. Weigelt, N. Wurzburger, L. M. York, S. A. Batterman, M. Gomes de Moraes, Š. Jane č ek,


1164 H. Lambers, V. Salmon, N. Tharayil, M. L. McCormack, A starting guide to root ecology:


1165 strengthening ecological concepts and standardising root classification, sampling, processing


1166 and trait measurements. _New Phytologist_ 232, 973–1122 (2021).


1167 89. C. P. Klingenberg, Size, shape, and form: concepts of allometry in geometric


1168 morphometrics. _Dev Genes Evol_ 226, 113–137 (2016).


1169 90. P. Pavlidis, J. D. Jensen, W. Stephan, A. Stamatakis, A critical assessment of


1170 storytelling: gene ontology categories and the importance of validating genomic scans. _Mol_


1171 _Biol Evol_ 29, 3237–3248 (2012).


1172 91. A. Korte, A. Farlow, The advantages and limitations of trait analysis with GWAS: a


1173 review. _Plant Methods_ 9, 29 (2013).


1174 92. J. Sampedro, E. R. Valdivia, P. Fraga, N. Iglesias, G. Revilla, I. Zarra, Soluble and


1175 membrane-bound β -glucosidases are involved in trimming the xyloglucan backbone. _Plant_


1176 _Physiol_ 173, 1017–1030 (2017).


1177 93. A. Bag, H. Krishna, P. . . Vinodh Kumar, N. B. Devate, R. Meena, S. Kumar, R. Patil,


1178 U. G. Reddy, J. B. Singh, N. Jain, P. K. Singh, G. P. Singh, Genetic mapping and candidate


1179 gene analysis of flag leaf and yield traits using a MAGIC population under heat stress in


1180 wheat (Triticum aestivum L.). _Plant Physiol. Rep._ 31, 331–346 (2026).


1181 94. B. Saatian, R. S. Austin, G. Tian, C. Chen, V. Nguyen, S. E. Kohalmi, D. Geelen, Y.


1182 Cui, Analysis of a novel mutant allele of GSL8 reveals its key roles in cytokinesis and


1183 symplastic trafficking in Arabidopsis. _BMC Plant Biol_ 18, 295 (2018).


1184 95. H. Lou, M. R. Tucker, N. J. Shirley, J. Lahnstein, X. Yang, C. Ma, J. Schwerdt, R.


1185 Fusi, R. A. Burton, L. R. Band, M. J. Bennett, V. Bulone, The cellulose synthase-like F3


1186 (CslF3) gene mediates cell wall polysaccharide synthesis and affects root growth and


1187 differentiation in barley. _The Plant Journal_ 110, 1681–1699 (2022).


45


