Objective/Meaning Taiwan Region in China has developed a compound fishery system featuring the coexistence of offshore, inshore and coastal fishing, as well as marine and inland aquaculture. There are obvious differences in terms of fishing fleet scale, aquaculture spatial layout, and the allocation of processing and cold-chain facilities among counties and cities. A unified accounting framework for fishery carbon emissions in counties and cities was constructed in this paper, and the role of production scale, industrial structure and production efficiency on carbon emissions was clarified, so as to identify the key links for emission reduction and clarify the low-carbon transformation path for the fishery industry in Taiwan.

Methods/Procedures  Based on the fishery statistics from counties and cities in Taiwan from 2013 to 2024, the carbon emissions from fishing and aquaculture production processes were measured by combining the activity data with emission factors. On the basis of describing the spatial and temporal characteristics of total carbon emissions, unit output emissions and unit output value emissions, the methods of two-way fixed effect model, extremely randomized trees and SHAP method were used to analyze the relationship between fishery production structure, production efficiency, industrial chain conditions and carbon emission intensity.

Results/Conclusions The study found that the overall carbon emissions from fisheries in Taiwan showed a fluctuating upward trend, with the fishing stage consistently dominating and emissions highly concentrated in a few major fishing and cities. The total amount of carbon emissions was not simply corresponding to the emission intensity, and some small-scale fishery counties and cities had higher emissions per unit output. There was a relatively stable positive correlation between the proportion of fishing production and carbon emissions per unit of output value. The efficiency of single-ship fishing and the efficiency of aquaculture per unit area were negatively correlated with carbon emission intensity. The effect of processing and cold chain conditions had a strong dependence on the accounting caliber. The results of machine learning further indicated that the proportion of fishing production was the primary predictor of the difference in carbon emission intensity among counties and cities, and its related impact exhibited significant non-linear characteristics. The research showed that the low-carbon transformation of fisheries in Taiwan should pay more attention to the optimization of production structure and the improvement of energy utilization efficiency.