2Background and aims 3 Upland rice is often grown where water and phosphorus (P) are limited and these two factors 4 interact on P bioavailability. To better understand this interaction, mechanistic models 5 representing small-scale nutrient gradients and water dynamics in the rhizosphere of full-grown 6 root systems are needed. 7 Methods 8Rice was grown in large columns using a P-deficient soil at three different P supplies in the 9 topsoil (deficient, suboptimal, non-limiting) in combination with two water regimes (field 10 capacity versus drying periods). Root architectural parameters and P uptake were determined. 11Using a multiscale model of water and nutrient uptake, in-silico experiments were conducted by 12 mimicking similar P and water treatments. First, 3D root systems were reconstructed by 13 calibrating an architecure model with observed phenological root data, such as nodal root 14 number, lateral types, interbranch distance, root diameters, and root biomass allocation along 15 depth. Secondly, the multiscale model was informed with these 3D root architectures and the 16 actual transpiration rates. Finally, water and P uptake were simulated. 17 Key results 18The plant P uptake increased over threefold by increasing P and water supply, and drying periods 19 reduced P uptake at high but not at low P supply. Root architecture was significantly affected by 20 the treatments. Without calibration, simulation results adequately predicted P uptake, including 21 the different effects of drying periods on P uptake at different P levels. However, P uptake was 22 underestimated under P deficiency, a process likely related to an underestimated affinity of P 23 uptake transporters in the roots. Both types of laterals (i.e. S-and L-type) are shown to be highly 24 important for both water and P uptake, and the relative contribution of each type depend on both 25 soil P availability and water dynamics. Key drivers in P uptake are growing root tips and the 26 distribution of laterals. 27 Conclusions 28 This model-data integration demonstrates how multiple co-occurring single root phene responses 29 to environmental stressors contribute to the development of a more efficient root system. Further 30 model improvements such as the use of Michaelis constants from buffered systems and the 31 inclusion of mycorrhizal infections and exudates are proposed. 32 33 34

23Upland rice production on weathered soils is often constrained by phosphorus (P) deficiency and soil 24 acidity. Farmyard manure application (FYM) can sharply enhance yields and agronomic P fertilizer (TSP) 25 efficiency. We tested the hypothesis that rice genotypes differ in the extent of using organic P and offering 26 distinct benefits under TSP-FYM combinations. Multiple field trials were conducted in the uplands of 27 Madagascar, with factorial combinations of six genotypes, FYM, and TSP applications, with blanket N&K 28 additions. Rice grain yields reached 6 t ha -1 after three years of TSP+FYM application, were lower when 29 FYM or TSP were used separately, while crops failed under zero P input. Genotypic differences were 30 inferior to the large treatment effects. Application of FYM increased soil pH and CaCl2-extractable P while 31 decreasing CaCl2-extractable aluminium. An additional liming trial indicated that beneficial effects of FYM 32 over TSP relate to liming effects. Genotypic ranking of yields and agronomic efficiency was inconsistent, 33 without superior genotypes under FYM versus TSP. However, Chomrong Dhan and FOFIFA 172 showed 34 superior yields under TSP+FYM. The FYM application lowers aluminium toxicity which overrules potential 35 effects of organic P supply. Aluminium tolerance should be included when developing rice genotypes for 36 low P tolerance in weathered soils.NA Data not available. a Phosphorus Acquisition Efficiency, expressed as total P uptake in mg at low P availability in the soil. b Root Efficiency, expressed as total P uptake per unit root size (either root biomass in g or root surface area in cm 2 ).