Estimação dos parâmetros populacionais de densidade e abundância para os golfinhos de rio da América do sul boto (inia spp.) e tucuxi (sotalia fluviatilis): aperfeiçoamento do método e abordagens ecológicas

Abstract

The boto (Inia spp.) and tucuxi (Sotalia fluviatilis) are freshwater small cetaceans endemic of South America. The river dolphins are among the species of cetaceans most threatened by growing and disorderly human activities, making it essential to know the population parameters for these species. Efforts to compute estimates of abundance for South American River dolphins have increased in the last several years and refinements of the methods employed to estimate population size are required. A mixed protocol of line and strip transects via Distance Sampling (DS) methods have been applied in the studies carried out with river dolphins in South America. In this study, we analyzed a 10- year dataset collected in 31 surveys through the Amazon, Orinoco and Tocantins- Araguaia River Basins for boto and tucuxi population estimates. Additionally, a distance calibration experiment allowed to infer about observer accuracy in sampling distances to the object detected. Through a GLM – Generalized Linear Models analysis, a slope of 0.952 (p<2e-16) shown high accuracy of distances sampled, there was no statistical difference in abundance estimates between estimated and real distances. Models with no covariates (Conventional Distance Sampling – CDS) and one or multiplex variables (Multi Covariate Distance Sampling – MCDS) were performed to evaluate best detection curve of detection function. MCDS methods were the best model for detection function of both species (Inia p = 0.39 (CV = 0.12), Sotalia p = 0.27 (CV = 0.20)), taking into account group size and sighting platform (bow and stern) as covariates. Using data for both sighting platforms, the detection probability at zero distance (g(0)) was estimated by Mark-Recapture Distance Sampling for boto 0.814 (CV = 0.053) and tucuxi 0.989 (CV = 0.006). Estimates of general detection function f(0) and detection probability g(0) to apply in samplings in different rivers has proved not to be the most accurate strategy. 4 When possible, f(0) and g(0) should be estimated as sampling-specific since biotic and abiotic factors, and hydro-geomorphology features directly influence in the parameters estimation. Hydro-geomorphology appears to acts as unit of distribution and population size of river dolphins in South America. Therefore, post-stratification analysis in subregions of the same river (Tocantins River) reduced by 70% the CV’s in the estimates. A relatively small population of boto was estimated to the lower-medium Tocantins River (736, CV = 0.52), and for Guaviare River (1138, CV = 0.32); otherwise, the Purus River were estimated 7672 boto (CV = 0.37) e 9238 tucuxi (CV = 0.49). Despite intrinsic features of river basins, several human activities at different levels, directly interferes in the interpretation of abundances estimates of each river. Refinements in analytical methods presented in this study increase the precision of results and can contribute to the improvement of population size estimates of boto and tucuxi in future studies.