Bioeconomia pesqueira em Pernambuco: análise econômica da produção e comercialização de pescado

Abstract

Fisheries and aquaculture have an essential role in the economics of several traditional communities from inland to the coastal zones. These activities have been highlighted in literature as relevant for addressing global problems associated with hunger and malnutrition, whether by generating income or by offering quality protein at affordable prices to low-income households. However, even though fishing and aquaculture play important socioeconomic roles for tropical, studies evaluating the economic dynamics of these sectors are scarce, and in the Brazilian case, generally limited to South and Southeast regions of the country. Given this, in the present work generalized linear models were analyzed to evaluate the economic aspects related to aquaculture and the fish market in the state of Pernambuco, with results divided into two works. In one of them, a census data survey was carried out with the aquaculture producers from Feira Nova municipality (Pernambuco’s Agreste), with the collection of information about management characteristics, aspects of the properties, cultures, and economic variables, used to calculate indices of farm’s profitability. Generalized linear models were then adjusted to understand the relationship between the variables and the calculated profitability indices. Subsequently, the Pernambuco Supply and Logistics Center (CEASA-PE) database was used to analyze the volatility of sardine prices sold in the state and evaluate the forecasting performance of two machine learning algorithms in the analysis of trade and commercialization of seafood items. In the first article, variables related to management and economy presented higher explanatory power for the profitability indices of aquaculture production in Feira Nova. However, there were important differences in how the variables are related to economic indices in shrimp and tilapia farming, evidencing the potential applicability of genearalized linear models in this analytical context. In the second article, the volatility of sardine prices varied throughout the period evaluated, showing peaks during the period of the new coronavirus (COVID-19) pandemic. Both evaluated machine learning algorithms demonstrated reduced error measurements, highlighting the applicability of these tools to forecast seafood prices, but the long short-term memory (LSTM) neural network performed better than Fbprophet. The results presented expand knowledge about the aquaculture economy and the price dynamics of Pernambuco’s fish market.