Geostatistical analysis and spatial modeling of population density for the sea cucumbers Isostichopus badionotus and Holothuria floridana on the Yucatan Peninsula, Mexico

Spatial distribution of Holothuria floridana and Isostichopus badionotus was examined at two sites off the coast of Yucatan: Isla Arena, Campeche, and Dzilam de Bravo, Yucatan, at different times using geostatistics. By means of ordinary kriging, inverse distance weighted and spatial simulation; seven models representing spatial distribution were produced and compared: three for H. floridana and four for I. badionotus. Spatial continuity of population density was calculated using an experimental variogram. Five types of models were tested (spherical, exponential, linear, linear to sill and Gaussian). Three statistics were obtained to aid in model output interpretation: the proportion of spatial structure, the regression coefficient, and the residual sums of squares for each variogram. Ordinary kriging produced the models with highest goodness-of-fit, indicating that population density was a regionalized variable, especially when populations are at unfished levels. H. floridana and I. badionotus displayed spatial patterns including patchiness, with some patches covering more than 60 km2. Fishing changed the spatial structure or patchiness degree of population density. Spatial structure of virgin stocks was better explained using exponential models; whereas spherical models better represented spatial structure after fishing impacts.