Search for a yield indicator for winter wheat – initial findings: Physiological plant traits of culm are the best ones for early indication

The requirements for yield security and food production safety are increasing under the changing climate and socio-economic conditions. Our research how extreme weather events impacts plant protection has also shown that, in addition to the useful “model” tools, there needs to be a way of characterizing the condition of crops in situ using valid yield indicators. A selection of yield-related indicator traits is a prerequisite for this. A yield indicator should represent the current state of a plant as early as possible and be easy to use under practical conditions. It is therefore used in a specific, unpredictable situation. Such special situations can occur especially under the influence of extreme weather conditions. A yield indicator can then be available as a tool of action. Requirements for such a yield indicator have been defined at the beginning of our investigations. In the search for such a yield indicator, we have therefore used biostatistical methods to examine historical data sets on important components of the yield formation process derived model functions for yield prediction using them. In our study, it was demonstrated that’s culm’s physiological parameters are suitable for assessing the current state of wheat plants, including their ability to compensate for abiotic and biotic stressors, and to forecast plants' yield. Promising candidates for the yield indicator of winter wheat are at BBCH 69/70: 1. Dry matter of the lower stem, 2. Dry matter of the entire stem, 3. (15)N-mass of the upper stem. Possible models for yield prediction are therefore: 1. y = 4.78+(19.64-4.78)/(1+e((3.)/ 0.39): ) dAIC 0.0 Pseudo-R² = 0.983; 2.:y=3.58+(19.11-3.58)/(1+e((6.75-x)/ 0.67)) dAIC 14.0, Pseudo-R² = 0.975; 3. y= Asym + (R0-Asym)*(e(-e(lrc) * x)) y= 19.23 + (-42.46-19.23)*(e(-e(0.77) * x)) dAIC 46.01, PseudoR² = 0.937. Successful indicator candidates and yield forecast models were tested in trials (2018-2023) with current winter wheat varieties. Successful candidates were validated and showed acceptable differences between forecasted and really yields. These findings may be taken as a basis for development of prognosis tools, e. g. applications for mobiles to predict yields under field conditions earlier, which work without any destruction of the plant or chemical analysis in the lab. The results are discussed in context with other modern methods for yield estimation and forecasting (i.e. remote sensing, vegetation indices, … already used in practice