ABSTRACT
We measured or estimated leaf and root physical and chemical traits of spatio-temporally heterogeneous field-grown sugar beet throughout its ontogeny during three growing seasons. The objective was to quantify the impact of temporal changes in these traits on root sugar content [S(R); g 100 g-1 root dry weight]. Artificial Neural Network (ANN), in conjunction with thermal time (ºCd), adequately delineated the boundaries (mean ± standard deviation, S.D.) between S(R) during early (41.6 ± 6.2), med (54.5 ± 3.0), and late ontogeny (63.4 ± 2.4), corresponding, respectively to low, medium, and high S(R). Calibration and validation Partial Least Squares (PLS) regression models, using plant physical and chemical traits, predicted and validated sugar content of sugar beet leaves [S(L)] and roots [S(R)] throughout its ontogeny with significant probabilities. Most physical and all chemical traits exhibited dynamic changes throughout plant ontogeny and, consequently, negatively or positively impacted S(R). The positive impact of S(L) and root volume (RV) on S(R) diminished towards the end of the growing season; whereas, the positive impact of root density (RD) and carbon:nitrogen (C:N) ratio in leaves [C:N(L)] and roots [C:N(R)] persisted throughout plant ontogeny. Specific leaf area (SLA), in particular, exhibited negative, then positive impact on S(R). The utility of physical and chemical traits of field-grown sugar beets in building reliable PLS models was confirmed using multivariate analysis on secondary statistics (residual mean square errors, RMSE and validation coefficients of determination, Q2) which discriminated between and correctly classified low (100%), medium (95%) and high (97%) S(R) groups. The findings may have implications to design management practices that can enhance C:N ratio and C-sequestration in roots, maintain optimum, but not excessive, N level in developing leaves and roots, optimize root sugar content and minimize its variation under field conditionss.
