Comparison of winter wheat (Triticum aestivum L.) root length density distribution models under salt stress

Abstract

Root length density (RLD) is an essential parameter for modeling crop growth and root uptake in soil-plant systems. However, it is challenging to measure and monitor under field conditions. Several RLD models contain many parameters that are built with the essential data obtained by some destructive sampling methods, such as the soil drilling and excavation methods, and thus they have limited application. Linear, polynomial, power, exponential and logarithmic models are commonly used to describe the RLD distribution, but it is not clear whether different models can simulate RLD profiles and the most suitable model for RLD distribution under soil salt stress. Therefore, the comparison of five winter wheat (Triticum aestivum L.) RLD distribution models under salt stress were conducted using RLD data that was obtained from the two field experiments in 2014 and 2015 based on the minirhizotron and digital image processing techniques, and a new simulation method, log-normal model was proposed in this study. The results showed that among the five common models, the exponential model had the highest coefficient of determination (R2) for RLD distribution simulation in soil, with the average R2 was 0.926, but the model could not accurately fit RLD in all soil depths, and it is especially meaningless above 10 cm soil layers. Compared with the five common models, the proposed log-normal model was the most suitable for fitting the observed RLD distributions under soil salt stress in different winter wheat growth with an average R2 of 0.986, and had a better describes the distribution of roots in soils. The results indicated that the proposed log-normal model was the most optimal for fitting the observed RLD distribution in saline soils.