Bayes theorem applied to the yield estimate of the Pacific sardine (Sardinops sagax coeruleus Girard) from Bahia Magdalena, Baja California Sur, Mexico

Abstract

 The yield of the Pacific sardine Sardinops sagax caeruleus from Bahía Magdalena, B.C.S., was analyzed using a stockrecruitment model. The model was stochastic, and it used the hypotheses of process error (H₁) in the model, and observation error (H₂) in the data. The results showed a positive bias in the management quantities and the parameters of the model. Confronting both hypotheses with a Monte Carlo simulation resulted in evidences of the effect of the observation error in the measurement of the adult stock of the sardine population. Statistical analysis supported in the Bayes theorem showed that the probabilities estimated from a maximum likelihood model for hypothesis H₁ are informative enough as prior probability. In this way, the maximun sustainable yield (MSY) of the fishery was 14,400 t with u_MSY = 0.35. The decision table showed that parameters of the model have a probability > 0.80 for α (density-independent parameter) between 0.040 and 0.058, while β (density-dependent parameter) varies between 1.6 and 2.2 with a probability > 0.85. The joint distribution of both parameters allowed a yield 10,100 t < MSY < 20,200 t per fishing season.