Yahui Hua, Ying Sun, Guangzhou Liu, Yunshan Yang, Xiaoxia Guo, Shaokun Li, Dan Hu, Wanmao Liu, Peng Hou
Maize (Zea mays L.) is an important food and silage crop with high production potential and economic benefits (Erenstein et al. 2022). Understanding the maize yield potential can provide reliable theoretical and practical support for achieving breakthroughs in grain yield (Meng et al. 2013). It was one of the effective ways to explore the grain yield potential based on the model on a national scale.
In China, maize was widely cultivated. The four major maize regions span from 21° to 53°N in latitude and 73° to 135°E in longitude and cover the most complex climatic conditions suitable for maize growing in the world (Xu et al. 2017). The regional climate heterogeneity tends to induce ecological adaptability responses of maize, which can also lead to variations in the interregional adaptability of crop models (Abbas et al. 2023). In addition, in the context of global climate change, crop production has been severely affected by extreme weather events, especially maize and wheat (Schmitt et al. 2022). Clarifying the regional adaptability of the crop growth models is an important basis for analyzing the response of maize to climate change.
Nowadays, crop growth models have been increasingly developed and applied, which play a crucial role in regional simulation, future climate scenario simulation, optimization of cultivation measures, and assessment of meteorological disaster impacts. Currently, the main crop models applied in China include AquaCrop, APSIM, and WOFOST models, most of them were mainly focused on wheat and rice and the application status varied with regions (Jin et al. 2016; Kheir et al. 2021). The Hybrid-Maize model is a maize-specific process model developed on the studies of high-yielding maize in the United States, which has been tested and widely used in the United States and South Asia (Yang et al. 2004; Timsina et al. 2010). It can simulate the maize grain yield potential in specific years and regions by inputting the required relevant parameters including meteorological data, the tested cultivars, and field management information (Yang et al. 2004). Most importantly, it provides a basis for clearing the maize grain yield improvement space and technical approach to reduce the yield gap. Currently, under the background of dense panting conditions to increase maize grain yield, the application of this model in China has gradually attracted attention. However, previous studies on the Hybrid-Maize model were mainly limited to specific regions or sites and applied under low planting conditions (Liu et al. 2012; Meng et al. 2013; Hou et al. 2014). There were few reports assessing its adaptability at large spatial scales and under dense planting conditions. The climatic conditions were diverse in different maize growing regions across China. Therefore, it is crucial to evaluate the adaptability of models under dense planting conditions in different regions. In this study, we evaluated the performance of the Hybrid-Maize model in the major maize growing regions of China based on field data at 22 experimental sites under high planting density (7.5×104 plants ha-1) during the period of 2011-2015. The maize growing information and climatic conditions of the experimental sites were listed in the Appendix A. The findings can provide a reference for the application of the Hybrid-Maize model under dense planting conditions in different regions of China.
The results of this study showed that the normalized root mean square error (NRMSE) were all below 30.0% and the index of agreement (D) were approached to 1, which indicated that the simulated yield was within an acceptable range (Fig. 1-A). Particularly, the model showed the best adaptability in the NW (Northwestern maize growing region, NRMSE=9.8%). A similar performance in the prediction of grain yield (NRMSE=7.1%) was observed in the United States under high density (Abimbola et al. 2022). It was mainly that the Hybrid-Maize model was developed based on high-yielding fields in the United States, where the planting densities commonly exceed 7.5×104 plants ha-1 (Yang et al. 2004). It was also shown that the average maize grain yield in the NW was significantly higher than that in the SW (Southwestern maize growing region), HHH (Huang-Huai-Hai maize growing region), and NM (Northern Spring maize growing region). The average measured grain yield in the NW (18.2 Mg ha-1) was higher than that in the SW (9.4 Mg ha-1), HHH (11.0 Mg ha-1), and NM (13.2 Mg ha-1).
Aboveground biomass and harvest index (HI) are the bases for maize grain yield formation. It was indicated that the model had better simulation effects in aboveground biomass with a lower NRMSE value in the NW (17.2%) than that of SW (24.8%), HHH (22.9%), and NM (26.2%) (Fig, 1-B). The simulation accuracy of this model for aboveground biomass in the NM (NRMSE=26.2%) was similar to a previous study conducted in the Northeast region under sufficient irrigation conditions (NRMSE=24.4%) (Liu et al. 2012). However, the performance of the model was slightly different in these growing stages with a similar trend in simulated accuracy (i.e., the jointing stage>silking stage>maturity stage) for these four regions. As for the harvest index, it was shown that the Hybrid-Maize model perform well in the NM and HHH under dense planting conditions (Table 1). The simulated HI for all regions with the trend of HHH>NM>NW>SW that differed from the spatial distribution (NW>SW>NM>HHH) in the previous study (Liu et al. 2020), which may be related to the overestimation of the HI in the HHH and NM and underestimation in the NW and SW by this model. Therefore, further optimization in the HI is required for this model when applied across different regions.
The dynamic changes in leaf area index (LAI) are shown in Fig. 1-C. There were significant differences in the performance of the Hybrid-Maize model between regions under dense planting conditions. The performance of the model in simulating LAI was better in the HHH (NRMSE=28.8%) and NM (NRMSE=22.0%) than that in the NW (NRMSE=33.4%) and SW (NRMSE=44.2%). Additionally, it was observed that the Hybrid-Maize model showed a good fitting degree in the jointing and mature stage period in most of these regions. Overall, this simulated values were higher than the measured with an average overestimation of 37.0% for the whole growth season. Specifically, the simulated values in the SW were on average 43.6% higher than the measured. In the NM, the simulated values were on average 12.5% lower than the measured after silking.
In summary, the Hybrid-Maize model showed good adaptability in the simulation of grain yield and dynamic changes of aboveground biomass in the four major maize growing regions of China under moderate planting density conditions, especially in the NW and SW. However, there was a significant underestimation of HI in the NW and SW. Conversely, there was an overestimation of LAI in these regions. Further evaluation of the model can be calibrated by adjusting the LAI and HI to refine the prediction of grain yield potential. Overall, the Hybrid-Maize model can provide relatively acceptable simulation references in the HHH and NM for all parameters under dense planting conditions.