Positive biodiversity-productivity relationships have been found in biodiversity field experiments of grassland, forestry, and other natural terrestrial ecosystems, where diversity effects were separated by complementarity (CE) and selection effects (SE). However, we know little about how CE and SE are related to root traits and root dissimilarity.

A four-year field experiment was carried out with a split-plot design, where main plot was four nitrogen (N) applications (N0, N1, N2, N3) and five cropping systems (maize (Zea mays L.)/soybean (Glycine max L. Merrill.), maize/peanut (Arachis hypogaea L.) intercropped and the corresponding monocultures) with three replicates. Roots were sampled in the N0 and N2 treatments in two years. Intercropping effects were analyzed based on grain yield for four years and roots were sampled down to 60 cm depth, and analyzed with morphological parameters at different crop growth stages in two years.

Intercropping significantly increased grain yield and aboveground biomass in both intercropping systems under all N treatments. The partitioning of the net intercropping effects showed that yield advantage in intercropping was due to a positive CE under the N0 treatment, and to a positive SE with N application.

Maize showed greater root morphological plasticity than the legumes did, with greater changes in root length density (RLD), root weight density (RWD) and total root surface (TS) in intercropping than in monoculture. Intercropped maize occupied a larger soil space, while lateral RLD distribution of legumes was decreased by maize. The RLD, RWD, and TS of intercropped maize were constant or increased in later growth stages. SE showed a significantly positive relationship with root dissimilarity. Principal component analysis showed mean root depth and specific root length of legumes drove the positive CE in the absence of N fertilization.

Root dissimilarity determined by maize explained the selection effects in overyielding. Complementarity effects under N0 were closely associated with specific root traits such as mean root depth and specific root length. Linking changes of root traits with intercropping effects aboveground helps understand yield advantages in diverse agroecosystem. In general, a cereal species with strong phenotypic plasticity intercropped with a legume species with strong physiological plasticity can maximize the yield advantage of intercropping.

This field experiment was arranged as a split-plot design with two factors from 2017 to 2020. The experimental factors included four N applications (N0, N1, N2 and N3) and five cropping systems, including maize (Zea mays L. cv. Xian-yu No.335)/soybean (Glycine max L. cv. Ji-yu No.47) intercropping, maize/peanut (Arachis hypogaea L. cv. Bai-sha No.1016) intercropping and the corresponding monocultures. The N application was set as the main-plot treatments, and the cropping system as the sub-plot treatments. The peanut monoculture as well as soybean monoculture received the same N-fertilizer applications, namely 0, 40, 80, and 120 kg N ha^-1. For the maize monoculture, these were 0, 180, 240, and 300 kg N ha^-1. The N applications for the maize/peanut intercrop were 0, 110, 160, and 210 kg N ha^-1, consistent with that in the maize/soybean intercropping system.

Each treatment had the same basal N application with four levels at 0, 40, 80, and 120 kg ha-1 N; the remaining N was equally applied twice at the jointing stage and pre-tasseling stage of maize. Each plot was 43.2 m2 (6 m×7.2 m) in size, with 12 rows (row distance was 0.6 m, plant distance was 0.17 m for maize and 0.085 m for legumes). Two rows of maize intercropped with two rows of legumes constituted one strip in intercropping plots . Maize and soybean were sown in early May, and peanut sown in mid-May; all crops were harvested in late September to early October.

The root samples were collected on July 20th in 2017 and 2018 (when maize was at pre-tasseling stage and legumes at blooming stage), and on August 15th in 2018 (when maize and legumes were at filling). The roots of maize and legumes were distinguished by differences in color, texture, and branching pattern. The roots of maize were white and smooth, and the roots of legumes were brown with small nodules. Separated root fractions were scanned using an Epson Perfection V750 Pro and then analyzed with Win-RHIZO™ giving data of total root length and total root surface (TS) for each soil core.