玉米生長曲線模型的應用與比較

Abstract

玉米 (Zea mays L.) 是重要的糧食，其產量以及進出口量在國際市場佔有重要地位。鑑於玉米本身的重要性，與產量密切相關的籽粒部分，便受到重視。因此了解玉米籽粒與相關的調查性狀，以及隨著時間推進，生長發育的情況，為本篇的研究重點。本研究使用甜玉米（臺農 5 號、華珍）及硬質玉米（臺農 1 號、臺農 7 號、明豐 3 號），收集春作與秋期間的莖桿、葉身、苞葉、穗軸以及籽粒乾物重。針對各品種地上部乾物重、苞葉、穗軸、籽粒乾物重以及果穗百分比，觀察在不同播種後天數時，其數值的變化，並獲得趨勢。並使用生長曲線模型 Logistic model 以及 Gompertz model，在應變數為籽粒乾物重，自變數為播種後天數的情況下，以及在應變數為果穗百分比，自變數為播種後天數的情況下，觀察生長曲線模型擬合的結果並進行模型比較，再運用 R2 觀察在應變數的變異中，可由自變數解釋的部分所占的比例，並運用 Akaike’s information criterion 進一步得到較好的模型。根據籽粒乾物重的模型擬合結果，甜玉米臺農 5 號在春作和秋作中，Logistic model 表現較佳，而硬質玉米品種則以 Gompertz model 較為適合。此外，果穗百分比的結果顯示，硬質玉米明豐 3 號在春作和秋作中，Gompertz model 的預測表現較好。期望本研究有助於了解各期作時各品種玉米時間上的變化，且藉由獲得較好的生長曲線模型，有助於了解玉米生長發育方面，隨著時間推移的情況。與前人玉米研究中關注的不同，本研究則著重於玉米籽粒及其相關數值的生長變化，期望為玉米的生長過程提供參考。由於本研究中僅比較 Logistic model 以及 Gompertz model 兩個生長曲線模型，未來或許能使用其他的生長曲線模型進行評估，並比較更多模型之間的差異。

Corn (Zea mays L.) is an important staple food of the world. The production, import and export of corn occupy a space in the world. In view of the importance of corn, the part of the kernel that is closely related to yield has received attention. This is the focus of this article in order to understand the corn kernel and related investigations, as well as their growth and development over time. In this study, we applied sweet corn (TNG5, Bright Jean) and flint corn (TNG1, TNG7, and MF3), with data collected on stalk dry weight, leaf blade dry weight, husk dry weight, cob dry weight, and kernel dry weight in spring and fall. We analyzed the aboveground dry weight, husk, cob, and kernel dry weight, and ear percentage of each corn variety through the changes at different days after sowing and obtain the pattern. By using growth curve model, logistic model and gompertz model, where the dependent variable is kernel dry weight, the independent variable is the number of days after sowing, and when the dependent variable is ear percentage, the independent variable is the number of days after sowing. Thus we tested the results in growth curve model for fitting and compare the simulation result among models. Then we used R2 to determine the proportion of the variation in the dependent variable that can be explained by the independent variables, and using Akaike’s information criterion to develop a further better model. The model The model fitting results for kernel dry weight indicated that the Logistic model was more suitable for sweet corn variety TNG5 in both spring and autumn seasons, while the Gompertz model was more appropriate for flint corn varieties. Additionally, for ear percentage, the Gompertz model provided better predictive performance for MF3 in both growing seasons. In this study, we expected it will help to understand the temporal changes of various varieties of corn during each crop season, and by our developing of a better growth curve model, it will help to understand the growth and development of corn over time. Different from previous corn studies that focused on traits such as leaf area, plant height, or total dry matter accumulation, this research emphasizes the growth changes of corn kernel and related traits, providing valuable insights into the corn growth process. Since only the logistic model and gompertz model were used in this study, these two growth curve models were compared. In the future, other growth curve models may be used for evaluation and the differences between more models can be compared.