Assessment Of Factors On The Wheat “Yield Gap” Of Dry Land With Long-term Fertilization

Wheat “yield gap” for dryland is the distance between the potential yield of rain-fed wheat and the actual wheat yield on the field,which is influenced by climatic factors,fertilizers,soil nutrients and wheat varieties and so on.It is difficult to properly evaluate the influencing factors on wheat “yield gap”,due to multicollinearity of these factors.In order to maintain the high yield and sustainable yield of wheat in the drylad of the Loess Plateau,and to solve the problem of appropriately evaluating the factors on wheat “yield gap”.This study was based on long-term experiment(starting from 1984),selecting six fertilizer treatments of continuous wheat planting and four fertilizers of wheat-pea rotation from it.The DSSAT model combined with field test method was used to simulate the potential yield of rain-fed wheat,and to evaluate wheat “yield gap”.At the same time,the temporal changes of meteorological factors,soil nutrients and soil water were analyzed,and analyze their influences on wheat “yield gap”.Partial least squares regression(PLSR)was used to eliminate the collinearity among factors and to identify the main controlling factors of the long-term continuous and wheat-pea rotation planting.It is of great significance to guide wheat production of dryland on the Loess Plateau,China.The main results and conclusions of this present study are as follows:(1)DSSAT model was suitable for application on the dryland of the Loess Plateau.The potential yield of rain-fed wheat on the Loess Plateau was 8324 kg·hm-2.The results of 31-year long-term experiment showed that wheat “yield gap” on the CK and P treatments were significantly higher than manure(M),nitrogen and phosphorus(NP)and NPM treatments on the wheat continuous cropping.The wheat “yield gap” on the P treatment was highest,which reached to 6996 kg·hm-2.Fertilizer contribution ratio(FCR)on the P treatment was lowest,which inhibition the FCR on the continuous cropping,reached to-18.3 kg·kg-1.The precipitation-use efficiency(PUE)was lowest on the P treatment,highest on the NPM treatment,which were 2.34 kg·mm-1·hm-2 and 7.45 kg·mm-1·hm-2.The wheat “yield gap” on the CK was highest,lowest on the NPM,on the wheat-pea rotation system.FCR and the precipitation-use efficiency(PUE)were highest on the NPM,lowest on the CK treatment.The wheat “yield gap” in the CK and single fertilizer treatment were higher than M and the combination of fertilizer treatment on the same precipitation pattern.FCR and PUE in the combination of fertilizer treatment were higher than single fertilizer treatment.The wheat “yield gap” in most of the treatments on the dry years were higher than normal years higher than wet years,where the most obvious effect were showed on NP and NPM treatments.(2)Meteorological factors in the long-term experiment research were found that accumulated daily mean temperature above 0°C during the wheat growing period(GDD),average temperature,total radiation solar(Rs),fallow-season precipitation(FSP)and annual precipitation(An P)(fallow-season precipitation+ growing-season precipitation)showed an increase trend among the 31 years.Growing-season precipitation(GSP)showed a decrease trend among the 31 years.Precipitation in March,May,June,October and December showed a decrease trend.While,precipitation in other months showed an increase trend among the 31 years.The correlation analysis between meteorological factors showed that An P and FSP had highly significant correlation under dry,wet and among the 31 years.The correlation between FSP and An P was 0.816,0.832 and 0.901,respectively,on the three types.The correlation between An P and September and December was 0.619 and-0.688,respectively under dry years.An P and August precipitation had a significantly positive correlation in the wet years.Wheat “yield gap” among FSP,An P and September precipitation had a significantly negative correlation under the different fertilizer treatments.Wheat “yield gap” and December precipitation had a significantly positive correlation on the different fertilizer treatments.(3)The analysis of soil nutrients in the long-term experiment found that soil organic matter(OM),total nitrogen(TN),available phosphorus(AP)and available potassium(AK)maintain a basic level under CK and P treatments and the interannual fluctuation range among them were smaller.By 2014,AP in the P treatment was 12.39 times higher than the starting of the experiment in the continuous cropping system.AK in the NP treatment was reduced by 9.58% than the beginning of the experiment.Soil nutrients in the NPM treatment were significantly higher than the beginning of the experiment.Wheat “yield gap” with AK had the significantly positive correlation on the CK.Wheat “yield gap” with soil nutrients had the significantly positive correlation on the P treatment.Wheat “yield gap” with soil nutrients had the significantly negative correlation on the other fertilizer treatments on the continuous cropping system.By 2014,AP in the P treatment was higher 12.90 times than the beginning of the experiment in the wheat-pea rotation system.AK in the NP treatment was reduced by 14.69% than the beginning of the experiment.Soil nutrients in the NPM treatment were significantly higher than the beginning of the experiment.Wheat “yield gap” had a significantly negative correlation among the soil nutrients under different fertilizer treatments.(4)The analysis of soil water in the long-term experiment found that annual averages of sowing water storage(SWS)and harvest water storage(HWS)were highest in the P treatment and lowest in the NPM treatment on the wheat continuous cropping.Absolute value of soil water consumption and wheat water consumption were the highest in the NP treatment and were the lowest in the P treatment.The absolute value of soil water consumption and wheat water consumption were 70.55 mm and 347.65 mm,respectively in the P treatment.Water-use-efficiency(WUE)in the combinative fertilizer treatments was significantly higher than that in CK and P treatments.The correlation analysis found that wheat “yield gap” with SWS and HWS had a negative correlation in the N and NPM treatment,positive correlation with soil water consumption.On the wheat-pea rotation system,the absolute value of soil water consumption and wheat water consumption were lowest in the CK treatment.The trend of SWS,HWS and PUE in the fertilizer treatments were similar in the continuous cropping system.Wheat “yield gap” with SWS and wheat water consumption had a negative correlation and positive correlation with soil water consumption,the correlation coefficient was-0.700,-0.817 and 0.826,respectively.(5)The multiple collinear diagnosis between wheat “yield gap” and precipitation factors showed that the precipitation factors had many multicollinearity.The high multiple collinearity problem was obvious between annual precipitation(An P)and fallow-season precipitation(FSP).Partial least squares regression(PLSR)analysis used can effectively solve the multiple collinearity between the independent variables.The PLSR model in the wheat continuous cropping system,wheat–pea rotation system and the combine with continuous and wheat-pea rotation systems showed that the first and second component in the three models had the large explanation on the wheat yield gap.The accumulation explanation of two components in three models were 66.7%,84.4% and 74.9%,respectively.In the first and second components of the PLSR model,nitrogen(N),phosphate(P)and organic fertilizer(M)played the dominated role.The analysis of the independent variable importance projection index(VIP)showed that N,P,M,FSP and An P played important role in the three models(VIP>1),and the effect of GDD,average temperature,Rs,wheat varieties,cropping systems on the wheat “yield gap” the three models were not significant(VIP<1).Through the regression coefficients the research,it was found that wheat “yield gap” was decreased with the increased of the amount of fertilizer,An P and FSP.