JIA-2019-11

2525 ZHOU Tian-mei et al. Journal of Integrative Agriculture 2019, 18(11): 2521–2533 for tomato (Fig. 1-C), and from 0.13 to 4.77 (MJ m –2 d –1 ) for cabbage (Fig. 1-D). When seeds were sowed on April 23, 2015, the corresponding TEP was the highest among the four sowing dates, indicating that the temperature and illumination during this period were relatively suitable for tomato and cabbage growth. Because the three fundamental temperature points for tomato and cabbage were different, the TEP differed even though they were sown on the same date. In general, when TEP was high, the composite indexes, specifically G value, HI1, HI2, and HI3 were high in general, but single indexes were not always high (Table 3). 3.2. Development of the simulation model for seedling quality indexes Development of the simulation model for tomato The relationships between root-shoot ratio, G value, and healthy indexes of tomato seedlings to TEP were all parabolas. The seedling indexes all increased with TEP, and showed non-linear relationships with TEP (Fig. 2). However, for tomatoes, the R 2 of the curve for TEP and root-shoot ratio was only 0.7584, suggesting that this curve fitting was of poor quality and the reliability of the model was low (Fig. 2-A). This result indicated that TEP is not ideal for generating fitting curves for the root-shoot ratio of tomatoes. The R 2 of the curve for TEP and G value was 0.9739 for tomato (Fig. 2-B), indicating that the curve represented robust correlation between simulated and observed G value and TEP. F- test analysis indicate that the model has high accuracy ( P <0.01), and independent data could be used for validation. The R 2 of the curves for TEP and healthy indexes were all high, especially for HI1 with the R 2 of 0.9772, suggesting that these healthy indexes are good for developing these growth simulation models (Fig. 2-C, D, and E). For curves of TEP and three healthy indexes (HI1, HI2 and HI3) in tomatoes, statistical F -test suggested that the models have high accuracy too. Because the models developed with G value and healthy indexes had high R 2 values, independent data sets were used to further validate them. Deve l opmen t o f t he s imu l a t i on mode l f o r cabbage Similar to growth models for tomato, the non-linear increase in root-shoot ratio, G value, and healthy indexes of cabbage seedlings as TEP increased also resulted in parabolic curves (Fig. 3). Poor R 2 value (0.7511) was also observed for the curve derived from TEP and root- shoot ratio; this low curve fitting degree indicated its less accuracy for predicting root-shoot ratio under a certain TEP. On the other hand, the R 2 of the curve for TEP and G value was 0.9595; comparable results were also obtained for the curves for the TEP and healthy indexes that have R 2 values of 0.97 or higher. Statistical F- test showed that the models have high accuracy at a level of P <0.01 (Fig. 3). Fig. 1 Changes in temperature (A), illumination (B) and daily thermal effectiveness and photosynthetically active radiation (DTEP ; C and D) during different sowing dates. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 Growing days (d) 2014/9/18 2014/9/28 2015/4/23 2015/5/12 0 2 4 6 8 Growing days (d) Growing days (d) Tomato 0 2 4 6 8 DTEP (MJ m –2 d –1 ) DTEP (MJ m –2 d –1 ) Temperature (°C) Illumination (J m –2 d –1 ) Cabbage A B C D 10 20 30 40 50 60 0 120 240 360 480 600 720 840 Growing hours (h) 2014/9/18 2014/9/28 2015/4/23 2015/5/12