Abstract
Rhizoboxes are soil-root compartments that may well provide the closest naturalistic conditions for studying root systems architectures (RSAs) in controlled environments. Rhizobox-based studies can however lead to mis-estimation of root traits due to poor recovery of roots and loss of fine root features during washing of roots. We used a novel scanner-based rhizobox system to evaluate: (i) RSA traits of Brassica rapa genotypes; (ii) the relationship between root traits recorded from rhizoboxes and those of harvested roots and (iii) genotypic variation of seedlings in response to external P ([P](ext)) availability. Brassica rapa genotypes were grown in soil-filled rhizoboxes abutting flatbed scanners and were watered once with either deionised water or a solution of 600 mu M KH2PO4 to approximately 80% field capacity on a weight basis. Shoot and root P concentrations ([P](shoot) and [P](root)) of the B. rapa lines grown on different [P](ext) were quantified. Visible root length at the surface of rhizoboxes constituted 85% of the total root length recovered from harvested root samples. High P supply induced a strong increase in [P](shoot) in all genotypes (P < 0.001) whereas low P supply generally led to greater partitioning to roots. Seed P concentration and tissue P concentration were correlated only at low [P](ext). Total root length was strongly correlated with tissue P content under both low [P](ext) (r = 0.81, P < 0.05) and high [P](ext) (r = 0.82, P < 0.05) conditions. The novel scanner-based rhizobox system used addresses the substantial limitations associated with current use of rhizoboxes to study root growth dynamics.