氣候變遷下臺灣龍眼產業之韌性評估

Abstract

氣候變遷所帶來的暖化和極端乾旱、暴雨事件，已影響全球多數地區的農業生產。龍眼是臺灣重要的果樹之一，廣泛種植於中南部中低海拔山區，有著粗放且產量大的特性。為因應氣候變遷對龍眼產業帶來的挑戰，理解氣候因子與龍眼生產之間的關聯性，以擬訂調適策略及管理計畫，對龍眼產業的永續發展相當重要。本研究以臺灣中南部三大龍眼產區為研究區域，並搜集相關農業統計、氣候和社會經濟資料，建立龍眼產量、價格預測與農戶生計之整合系統動態模型。首先利用氣候因子模擬龍眼產量，再藉由供需法則推估市場價格，最後綜合各類社會經濟因子預測農家經濟變化，以評估氣候變遷下臺灣龍眼產業之韌性。

研究結果顯示龍眼前一年產量及冬季營養梢生長期氣溫為龍眼當年度產量之重要影響因子，其次為營養梢生長期雨量和開花期雨量。比對龍眼歷史產量資料，模型模擬結果的平均誤差（ME）、平均絕對誤差（MAE）、均方根誤差（RMSE）及平均絕對百分比偏差（MAPE）分別為-371.1 kg/ha、1526.4 kg/ha、2131.9 kg/ha及28.7%。於未來氣候變遷情境下，本研究發現無論是SSP1-2.6、SSP2-4.5、SSP3-7.0或SSP4-8.5情境下，全臺灣龍眼單位產量(kg/ha)都會下降，整體平均產量約下降30.9%，越南方的產區產量減損狀況越明顯。農家經濟推估結果顯示，高雄、臺南溪南地區的龍眼產業於未來氣候變遷下將難以達到經濟永續性，其它產區則恐面臨比過往更高的風險。

本研究建立之整合模型，可用來分析作物生長、天然災害、產地價格、社會經濟等層面對整體龍眼產業所造成的影響，以及在氣候變遷下所面臨的風險，可供農業從業人員和相關決策制定者參考。

The warming and extreme events of drought and heavy rainfall caused by climate change are negatively affecting agricultural production in many regions around the world. Longan, which is one of the important fruit trees in Taiwan, is widely cultivated in the middle and low-altitude mountainous areas of central and southern Taiwan, known for its extensive cultivation and high yield. To cope with the challenges posed by climate change, it is crucial to understand the relationship between climate factors and longan production. This understanding can help formulate adaptation strategies and management plans for the sustainable development of the longan industry.

This study focused on the three major longan production areas in central and southern Taiwan. Agricultural statistics, climate data, and socio-economic information were collected to establish an integrated system dynamic model for predicting longan yield, prices, and farmers' livelihoods. This model first simulated longan yield using climate factors, and then estimated market prices using law of supply and demand. Finally, various socio-economic factors were integrated to predict changes of farmers' livelihoods, evaluating the resilience of Taiwan's longan industry under climate change.

The research results showed that for longan production, the previous year's longan yield and the winter temperature for the stage of shoot development were the most important factors affecting the longan yield, followed by the rainfall during the stage of shoot development and flowering. Based on historical data testing, the average error (ME), mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) of the model simulation results were -371.1 kg/ha, 1526.4 kg/ha, 2131.9 kg/ha, and 28.7% respectively. This study found that regardless of the SSP1-2.6, SSP2-4.5, SSP3-7.0, or SSP4-8.5 scenarios, longan yield per hectare in Taiwan would decrease, with a mean 30.9% lower than historical data, and the decline in southern was even more pronounced. The economic estimation results showed that the longan industry in Kaohsiung and southern Tainan area was not economically sustainable, while other production areas faced higher risks compared to the past.

The integrated model developed in this study can be used to analyze the overall impact of crop growth, natural disasters, supply prices, socio-economic factors, and the risks faced by the longan industry under climate change. It can serve as a reference for agricultural practitioners and decision-makers.