ABSTRACT
Structural dimensions, digitally measured on stems and leaves of soybean plants during the first six reproductive growth stages (R1-R6), were used to assess the impact of five management strategies (combinations of cropping systems, tillage practices and crop rotations) on grain yield per plant. Stem and leaf dimensions, light penetration within the canopy [log (I/Io) x 100], fractal dimension of plant skeletal images (Do) multiplied by leaf area index (LAI), and midday differential canopy temperature (dT) explained 84.0% of variation among plant samples grown under five different management strategies, with 75-100% correct classification. Management strategies, growth stages and their interaction accounted for a total of 24-79% of variation in different structural dimensions and for 97%, 97% and 94% of variation in dT, LAI*Do, and log[(I/Io) x 100], respectively. Grain yield per plant can be predicted at R3, R4, R5 and R6 with increasing probability (R2= 58, 64, 69 and 72%, respectively) while decreasing root mean square error of the validation models (from 2.33 at R3 to 2.1 g per plant at R6) using dT, LAI*Do, and [log(I/Io) x 100] as predictors.
