Monitoring And Diagnosing Leaf Senescence Under Heat Stress After Anthesis With Hyperspectral Reflectance In Wheat

Warmer climate is the trend of future climate change.It increases the average temperature and brings more extreme temperature events,such as heat waves.Heat stress is one of the major abiotic adversities in wheat(Triticum aestivum L.)production.It causes the losses of yield and quality due to premature senescence,especially during the reproductive period.Thus,timely monitoring of senescence in wheat leaf and detection of premature senescence of plant,which is caused by heat stress,are critical for wheat production,especially under stress conditions.Many biological parameters have the ability to present the leaf senescence process,yet few researchers have systematically compared the differences between these parameters in their detecting efficiency.In this study,we analyzed the sensitivity of six biological parameters[the ratio of green area in leaf(GAR),the leaf chlorophyll content(LCC),the net photosynthesis rate(Pn),the maximum photochemical efficient(Fv/Fm),the equivalent water thickness(EWT)and the specific leaf weight(SLW)]to the senescence of wheat leaf.The results suggested that LCC,Pn and Fv/Fm were more efficient in characterizing the different processes of leaf senescence than did the other parameters.We concluded that LCC,Pn and Fv/Fm were the senescence-sensitive biological parameters,and they had a potential to monitor the leaf senescence and to detect the premature senescence.Based on the senescence-sensitive biological parameters,we assessed the relationships between spectral indices(17 published vegetation indices and new two-band spectral indices)and the three senescence-sensitive biological parameters(LCC,Pn and Fv/Fm)to identify the spectral indices that are more senescence-sensitive.The results showed that the chlorophyll index-red edge(CIred-dge),normalized difference red edge index(NDRE)andphotochemical reflectance index(PRI)had the strongest correlations with LCC,Pn and Fv/Fm,respectively.In calibration,the determination coefficients(R2)of these three indices were 0.71,0.60 and 0.51,respectively;while in validation,the NRMSE of the three indices were 0.14,0.26 and 0.15 respectively.As the CIred-edge and NDRE were similar and strongly related to both LCC and Pn,we decided that NDRE and PRI were the leaf-senescence-sensitive spectral indices.Additionally,we explored the suitable bandwidths for these optimal spectral indices.The results indicated that the suitable bandwidth for the visible and red edge bands of sensitive spectral indices was 15 nm,and the near infared bands were not evidently affected by the bandwidths.Heat stress is one of the major causes of premature senescence in wheat.Real-time monitoring of heat stress is important in predicting premature senescence.Among these three senescence-sensitive biological parameters(LCC,Fv/Fm and Pn),Fv/Fm and Pn were able to more efficiently monitor the start of heat stress and the end of relative lesser stress,but LCC showed lower performance here.Among the three senescence-sensitive spectral indices,PRI was more effective to in monitoring the heat stress.Nevertheless,the sensitivity of PRI was less than that of Fv/Fm and Pn,and it failed to detect the beginning and end of heat stress lasting for three days.The ability of PRI to detect heat stress became similar to that of Fv/Fm when the duration of heat stress was increased to seven days.In conclusion,Fv/Fm is the optimum indicator for detecting early-stage heat stress,in which only the photosynthetic functions change.In contrast,PRI,a non-destructive indicator,works well to indicate relatively late-stage heat stress,in which the chemical and physical characteristics of leaves(e.g.,chlorophyll content)are affected.The major objective of the present investigation was to diagnose the wheat premature senescence under heat stress.We analyzed the relationships between the dynamic characteristics[the accumulated difference(AD),the relative value(RV)and change rate per day(CR)]of the biological parameters and the premature senescence degree[ratio of reduced growth duration after anthesis(RGD)].The results showed that there existed strong correlation between the AD of LCC and Pn(AD,cc and ADPn)and the RGD.It meant that it has was potential to use the ADLcc and ADPn to detect the premature senescence in wheat.We also used the dynamic characteristics of the spectral indices[the normalized difference vegetation index(NDVI),the NDRE and the PRI]to detect the wheat premature senescence.The results showed that there was also a strong correlation between the AD of PRI(ADPRI)and the RGD.It meant that the ADPRIu also had the potential in detecting premature senescence in wheat.In conclusion,the dinogis models of indicators were:(1)RGD?(-1.95 × DAA?23.56)× ADLcc+(-0.045 × DAA+1.75),(2)RGD=(-0.13 × DAA+1.70)× ADPn+(-0.079 × DAA+2.03)and(3)RGD=(7.5 × DAA+22)× 10-3 × ADPRI+(4.5×DAA-18)×10-3.Leaf area index(LAI)is an important variable for demonstration of wheat senescence.In this study,we assessed the performance of 16 published spectral indices,the new two-band spectral indices(NDSI)and the three-band spectral indices(mNDSI)in leaf area index(LAI)estimation.The results suggested that the performance of the two-band and three-band spectral indices were better than that in of the published VIs.Between these two kinds categories of new spectral indices,the two-band spectral index NDSI(P940,p730)was efficient to in estimating the LAI in wheat,while the three-band spectral index mNDSI(?940,p730,p950,0.8)was less efficient in under the condition that the red edge band had been contained in the two-band spectral indices.We concluded here that NDSI(?940??730),which contained the red edge band,was the optimal spectral index for LAI prediction and had the potential for identification of senescence in wheat.