Sources of sampling bias in enclosure fish trapping: Effects on estimates of density and diversity

Enclosure traps have been increasingly used to sample fish populations in shallow aquatic habitats because they have proven to be practical, effective and sufficiently precise. However, they are somewhat biased, usually underestimating fish densities. Large fish are under-sampled, due either to escape or scarcity, consequently affecting biomass estimates. Using computer simulations and an extensive enclosure-trap data base, we found that the most consistent explanation for the reduced densities observed was a reduced effective sampling area, rather than an inappropriate sampling model. Corrections for density bias could be made by considering that a 1-m² trap effectively samples a 0.81-m² area of marsh. Using an index of species diversity, we confirmed that, despite the bias in density estimates, the composition of the fish community could be accurately described using data obtained from enclosure trapping. Enclosure trapping could effectively be argued to be either with- or without-replacement sampling. Thus, under field conditions, fish may be either removed or replaced without violating the sampling model.