比較不同模式耦合溫度變化與農地變異模擬結球萵苣的生長

Abstract

結球型萵苣 (Lactuca sativa L.) 是台灣秋冬季出口的指標蔬菜作物，由於植株結球的表徵特性易受夏季高溫限制，所以在台灣的主要栽培季節固定從秋末至隔年晚春；就植株生長過程，已展開外葉是光合產物的主要供源，其葉球是積儲也是生產標的，外葉生長表現將是主導葉球產量的首要栽培管理指標。線性 (LIN)、Gopertz (GOP) 和Logistic (LOG) 模式是使用最普遍的生長模式，因應不同作物生長特性之差異，生長模式的最佳化是提升智慧農業的作物栽培管理效益所不可或缺。本研究依據相對葉生長率 (relative leaf-growth rate; RLR) 和相對生長速率 (relative growth rate; RGR) 之定義為建構結球萵苣生長模式的理論基礎，以LIN、GOP和LOG 模式套配地上部乾重和外葉面積的生長曲線；並考慮產期溫度變化和農地變異，提出模式耦合程序改善不同期作溫度差異和不同田區環境差異造成的模擬預測偏差。本研究的試驗區位於雲林縣二崙鄉和麥寮鄉，試驗田區地號包括：二崙鄉油車段427 號和公館段382 號、麥寮鄉山寮段474 號和550 號；於2017 年和2018 年在427 號與382 號田區各進行1 期試驗，於2017 年在474 號和550 號田區各進行1 期試驗，共彙集6 組結球萵苣的試驗資料供模式套配評估與驗證。結果顯示，LIN、GOP 和LOG 模式皆適用於描述外葉面積和地上部乾重的生長曲線，但各模式的適配性可能因期作氣候及田區環境差異的影響而有差異，套配模式估算的參數a 和b 不僅可供說明生育期間RGR 和RLR 的變動趨勢，而且可用於溫度耦合方程式a(T) 和b(T)評估季節性溫度差異對植株生長速率的影響，並估算出耦合參數a' 和b'；再由溫度耦合方程式結合特定田區(S) 的植株生長資料，可估算出耦合參數c'。依據3 種模式的定義代入耦合參數值：a'、b' 及c'，即可獲得各模式同時耦合溫度變化及農地變異的生長模擬函數：FS(t | a', b', c')LIN、FS(t | a', b', c')GOP 和FS(t | a', b', c')LOG。耦合模擬結果顯示，相較LIN 和GOP 模式，以LOG 模式耦合模擬外葉面積和地上部乾重在生育期後段的預測準確度高，較有利準確推估最適採收期與最終收穫產量，若以耦合程序的操作及應用性考量，以普遍適用的LIN 和GOP 模式可被優先選擇。 Iceberg lettuce (Lactuca sativa L.) grown during the cool seasons in Taiwan is the flagship vegetable to export,. High temperatures in summer will retard or violate the heading physiology of iceberg lettuce, so Taiwan’s iceberg lettuce is routinely grown from early autumn to the late spring of next year. In the growth of iceberg lettuce, the outer leaves expending would be a major source of photosynthates; the leafy head which accumulates abundant photosynthates would be a sink. The growth of outer leaves will govern the yield of the leafy head. In smart agriculture, it is essential to optimize the growth model to increase the benefits of cropping management in smart agriculture, while coping with the growth variation in different crops. Linear (LIN), Gopertz (GOP), and Logistic (LOG) models are frequently used for modeling crop growth. In the present study, relative leafgrowth rate (RLR) and relative growth rate (RGR) are fundamental to developing the growth model of iceberg lettuce. The growth curves for shoot dry weight and outer leaf area were fitted to LIN, GOP, and LOG models. And model coupling procedures with the seasonal temperature changes and farmland variation were proposed to improve the prediction of the growth modeling. Two experimental sites, No. 427 and 382, are located at Erlun Township, Yunlin County and the other two sites, No. 474 and 550, are located at Mailio Township, Yunlin County. The growth survey of lettuce plants at sites No. 427 and 382 was carried out in the Winter 2017 and Spring 2018 cropping seasons, respectively. At sites No. 474 and 550, the growth survey of lettuce plants was only conducted in the cropping season Winter 2017. A total of six datasets were used in the study for model fitting assessment and validation. The results showed that LIN, GOP, and LOG models were well fitted to the growth curves of the outer leaf area and shoot dry weight. The goodness-of-fit for the models would be varied by the climate of growing seasons and environmental variation of sites. Parameters a and b obtained from the models fitting could be used to describe the dynamics of RGR and RLR in growth periods; also, the values of a and b were imported by the coupling expressions, a(T) and b(T), with temperature (T) to evaluate the influences of seasonal temperature changes on the growth rates. In addition, a(T) and b(T) were combined with data of plant growth recordings on the site (S) to obtain coupling parameters a', b', and c'. Then, the growth modeling functions, FS(t|a', b', c')LIN, FS(t|a', b', c')GOP, and FS(t|a', b', c')LOG, corresponding to LIN, GOP, and LOG models were developed, respectively. Compared with FS(t|a', b', c')LIN and FS(t|a', b', c')GOP, FS(t|a', b', c')LOG was more accurate in the predictions of outer leaf area and shoot dry weight at the later growth periods; thus, LOG model used in the coupled modeling procedure would be more suitable for prediction of the time to harvest and the yield of leafy heads. If the range of application and convenience of use are the priorities for modeling the growth of iceberg lettuce, LIN, and GOP models will be the superior options in models coupling.