Nitrogen Revising of Rapeseed (Brassica napus L.) Phenology and Leaf Number Models

The Decision-making System for Rapeseed Optimization-Digital Cultivation Based on Simulation Models, DSRODCBSM, is a dynamic model that describes the growth and development of winter rapeseed. In order to perfect rapeseed growth models, Ningyou16 (NY16), Ningyou 18 (NY18), and Ningza 19 (NZ19) were adopted as materials, and the field experiments with 2 cultivars and 2 nitrogen levels, and pot experiment with 3 cultivars and 2 nitrogen levels were conducted during 2007-2008, 2008-2009, and 2011-2012 in Nanjing, respectively. The experimental results showed that the phenology and leaf number in rapeseed models had obvious difference for the same cultivars under different nitrogen levels. Thus, the nitrogen effect factor, F (N), was put forward, used in the phenology sub-model in rapeseed growth models, and the verification of the leaf number sub-model can be done through model parameter adjusting. The simulated values before and after using F (N) and the observed values were compared, and the precision for the phenology sub-models in rapeseed growth models were raised further.


Introduction
Rapeseed is one of very important oilseed crops in the world, and its plant area in normal year is about 18-30 million ha. The plant area of rapeseed in China is about 6-7 million ha, and its total yields is about 10-13 million tons, which ranks the fifth place in crop production in China [1]. It plays a very significant role in ensuring cooking oil and plant protein supply, and promoting farmer income increase that makes rapeseed production stable sustainable growth. However, the good cultivars and the relevant advanced management techniques are very important to promote rapeseed production with high yield, good quality, high benefit, ecology, and safety. In that the rapeseed growth models is an important basis of rapeseed precision management techniques.
In recent years, studies on rapeseed crop models have made rapid progress. Notably, some rapeseed growth and development models, and ecological system models, e.g. EPR95 (erosion-productivity influence calculator, EPIC-Rape) [2], DAR95 (differential algebra for identifiability of systems, DAISY-Rape) [3], LINTUL-BRASNAP (light interception and utilization simulator) [4], CERES-rape (crop environment resource synthesis) [5], APSIM-Canola (agricultural production systems simulator) [6], and CECOL [7], etc. had been developed which can simulate rapeseed growth and development in real time. In China, the research on rapeseed growth model was not more. Liu and Jin [8], and Liu et al. [9] set up rapeseed phenology model etc. Zhang et al. [10], Cao et al. [11][12][13][14], and Tang et al. [15,16] studied the rapeseed growth and development simulation models, optimization models for rapeseed cultivation, and soil moisture and nitrogen dynamic models during rapeseed growth season, and the Decision-making System for Rapeseed Optimization-Digital Cultivation Based on Simulation Models (DSRODCBSM) were developed combining the rapeseed growth models (including phenology, leaf number, biomass, leaf area index (LAI), and shoot number dynamic models, etc.), the rapeseed optimization models (including the optimum season, the optimum LAI, the optimum shoot numbers, the optimum sowing rate, the optimum fertilization rate, and the optimum soil moisture, etc.), and expert knowledge of rapeseed plant diseases and insect pests, based on field experiments in Yangtz river middle valley of China [11], employing ideas of Rice or Wheat Cultivation-Simulation-Optimization-Decision making System (R/WCSODS) [17,18]. However, the rapeseed phenological models, and the leaf number models in DSRODCBSM were established under the optimum soil nitrogen, and water conditions, etc., if they were used in different soil nitrogen, and water conditions, there must be some errors in their results.
The objectives of this study were to introduce the effect factor of nitrogen in the phenology and leaf number sub-model (APPENDIX A, and B) in rapeseed growth models based on the field and pot experiments during 2007-2008, 2008-2009, and 2011-2012 in Nanjing, test, and perfect rapeseed growth models.
The soil type of the experimental area is a hydragric anthrosol. Soil test results indicated the following: organic carbon, 13.7 g kg -1 ; total nitrogen, 54.95 g kg -1 ; available phosphorus, 24.25 g kg -1 ; available potassium, 105.03 g kg -1 ; and pH, 7.84.

4
The phenophase, LAI, the total shoot numbers, dry matter, leaf number, leaf photosynthesis, plant characters, and soil data, etc. were observed during rapeseed growth or after harvest.
The meteorological data during the experiments were down from Center for China Meteorological Information of China Meteorological Bureau.

Data Process
In this study, Excel.2007 and SPSS V 16.0 were used to analysis experimental data. The experiment data in 2008-2009 were applied to model establishment and parameter determination, and the experiment data in 2007-2008, and 2011-2012 were applied to model verification.

Model Verification
Simulation values were calculated in DSRODCBSM, and model precision was verified using root mean squared error (RMSE), mean absolute error (d a ), the ratio of d a to the mean observation (d ap ) [19], the determined coefficient (R 2 ), and 1:1 plotting between measured values and simulated values. If da and RMSE were smaller and R 2 was larger, the simulated values were better agree with measured values, i.e. the deviation between simulated values and measured values was smaller, and simulation results of model were more accurate and reliable. The calculation formula of RMSE and da can be expressed as follows: where X Oi is observed values, X Si is simulated values, d a is absolute error, |d a | is a absolute value of d a , d ap is the ratio of d a to the mean observation, and n is sample numbers.

The phenology and leaf number under the different nitrogen rate
3.1.1 Phenology. Under the local normal sowing date in 2007-2008, the phenology of different nitrogen levels for same cultivars had obvious difference at enlongation, and the enlongation date under N application conditions were later than that of CK. But the mature dates were not difference (Table 1). Under the local late sowing date in 2011-2012, the phenology of different nitrogen levels for same cultivars had obvious difference at mature date, and the mature date under N application conditions were later than that of CK (Table 2).  (Table 3, and 4), and the leaf number in main stem of different nitrogen levels for NY18, and NZ19 had no obvious difference (Table 3, and 4).   Fig.1, and Fig. 2, and the results showed that the nitrogen content in leaf had a peak value at pre-over-wintering (8 JAN 2008) under nitrogen application conditions, in contrast, had a vale value at the same time under CK conditions (Fig. 1); the nitrogen content in silique had a peak value around end anthesis under nitrogen application conditions, in contrast, had a vale value at the same time under CK conditions (Fig. 2). It set a basis for developing the effect factor of nitrogen in the next step. (1) where TRN is the actual leaf nitrogen content (g kg -1 ) around 10 d after fertilizing at pre-over-wintering, TLN is the lowest leaf nitrogen content (g kg -1 ) in the same time for CK, and TCN is the critical leaf nitrogen content (g kg -1 ). In that TLN and TCN can be obtained using the experiment data in 2007-2008, taking TLN=9.58 g kg -1 (Fig.1) for CK at pre-over-wintering, and TCN=9.88 g kg -1 . Due to the effects of nitrogen application on leaf numbers in main stem were different with cultivars, and years, the leaf numbers in main stem in sub-model can be verified through adjusting cultivar parameters in leaf number sub-model.  (Table 5). We can see that kj, basic development coefficient which was determined by cultivar heredity, was different for various cultivars in the same development stages apart from stage Ⅱ (emergence to vernalization), and pj (the genotypic coefficient of temperature effects for increasing), qj (the genotypic coefficient of temperature effects for decreasing), and Gj (the genotypic coefficient of photoperiod effects) were the same for various cultivars in the same development stages. Note: j represents development stage I (planting to emergence), Ⅱ(emergence to vernalization), Ⅲ

3.3.2
The validation of the phenology sub-model after revising. The comparison between phenology with F (N) and no F (N) were shown in Table 6, and Table 7, Fig.3, and Fig.4. The results showed that the phenology with F (N) were more close to the observed values of nitrogen treatments (Table 1, and Table 2 The phenology and leaf number in rapeseed were affected by multi-factors, which were decided by genotypes and environmental factors, and temperature and light in environmental factors were main factors of them. In addition, the phenology was also affected by fertilizer, water, and so on. However, the nitrogen impact factor, F(N), was only introduced into the phenology model in this paper. The water impact factor should be considered in future studies. As to the relationship between leaf number in main stem in rapeseed and nitrogen application, it was different under various cultivars, and needed to be studied further. 4.2 The effect factor of nitrogen. It was determined according to the changes in nitrogen content in leaf and silique of various cultivars and nitrogen rates (Fig. 1, and Fig. 2), and because leaf nitrogen content, TRN, can be acquired easily comparing with silique, the actual leaf nitrogen content around 10 d after fertilizing at pre-over-wintering was introduced to the effect factor of nitrogen. 4.3 The phenology difference of different nitrogen levels for various years. Why difference of the phenology of different nitrogen levels during 2007-2008 was at enlongation, while that of during 2011-2012 was at mature, the reason maybe was from higher month average temperature, and lower month rainfall in that time comparing with the normal year, e.g., MAR 2008 (the month average temperature was higher than the normal year at 2.6℃, but the rainfall was lower than the normal year at 49.6 mm), MAY 2012 (the month average temperature was higher than the normal year at 1.4℃, but the rainfall was lower than the normal year at 39.7 mm) ( Table 8 and Table 9). Note: NY, AT, ANT, AXT, ST, and R denoted the normal year, average temperature, average min temperature, average max temperature, sun times, and rainfall, respectively. The same as Table 8.

Coclusions
This paper presented an attempt at validating and perfecting of phenology, and leaf number sub-model in rapeseed growth models. Through the 3 year field experiment data analysis, we can conclude that the phenology in rapeseed models had obvious difference for the same cultivar under different nitrogen levels. Thus, the nitrogen effect factors were put forward and used in the phenology sub-model in rapeseed growth models, and the verification of the leaf number sub-model can be done through model parameter adjusting.
The simulated values before and after using nitrogen effect factors and the observed values were compared, and the precision for the phenology sub-models with nitrogen effect factors in rapeseed growth models were raised further.

A. Phenology
The basic models of rapeseed phenology were developed in the thesis through employing ideal of -Rice Clock Models‖ [11][12][13][14] [17][18]. dP j /dt = 1/D Sj = e kj ·( T ebj ) pj · (T euj ) qj · (P ej ) Gj · f(E Ci ) T ebj = (T i -T bj ) / (T oj -T bj ), when T i <T bj , T i = T bj ; when T i >T oj , T i =T oj T euj = (T uj -T i ) / (T uj -T oj ), when T i >T uj , T i = T uj P ej = (P i -P bj ) / (P oj -P bj ), when P i <P bj , P i = P bj ; when P i >P oj , P i =P oj where dP j /dt is the development rate at the j th stages, D Sj is the days at the j th stages, T ebj and T euj are the effective factors for temperature, respectively, kj is basic development parameter which is determined by cultivar heredity, pj and qj are the genotypic coefficient of temperature effects, P ej is the effective factor of photoperiod, Gj is the genotypic coefficient of photoperiod effects, and f(E Ci ) is the effective function of agronomic practice factors for rapeseed, T i is the daily mean temperature (℃) in the j th stage, T bj , T oj and T uj are lower, optimum, and upper limit temperature (℃) demanded in the j th stage for rapeseed, respectively, and P bj , P oj are the critical and optimum day length (h) demanded in j th stage for rapeseed, respectively. Vernalization models can be described as following through employing ideals of -wheat clock models‖: dV/dt = 1 / D s2 = e k2 ·( V e ) C If a cultivar was winter or semi-winter rapeseed, the expression of V e was: However, if it was spring rapeseed, the expression of V e was: where K2 and C are the parameters of vernalization, V e is the factor of rapeseed vernalization effect, V ti is the daily mean temperature in vernalization phase. It will finish vernalization phase when V e equal to some extent accumulation days; the vernalization days of the winter rapeseed were 30 to 40 days, the semi-winter rapeseed with 20 to 30 days, and the spring rapeseed with 15 to 20 days.

B. Leaf number
The growth rate of rapeseed leaf were different in different varieties, development stages, temperature, and nutrition conditions etc., when nutrition condition was optimum, the models of rapeseed leaf number were [11][12][13][14] [17][18] : dL j /dt = f (L j ) = 1/D Lj = D Loj ·( T t /T o ) La/Lb D Loj = e LK where dL j /dt is the development rate of the j th leaf , f (L j ) is the basic development function, D Lj is the development days demanded from emergence to the j th leaf number, D Loj is the development days demanded from emergence to the j th leaf number under the optimum conditions, T t and T o are the daily mean temperature (℃) of the t th day, and the optimum temperature for rapeseed leaf number development, respectively, and La, Lb, and LK are the parameters of leaf models, respectively.