FIELD TRIALS AND SIMULATIONS OF POINT-NEAREST-NEIGHBOR DISTANCE METHODS FOR ESTIMATING ABALONE DENSITY

We investigated evidence for bias in estimates of abalone density from the point-nearest-neighbor (PNN) diver survey method wherein divers measure distances between abalone and from random points to nearest abalone. Field and simulation tests of the PNN survey method were undertaken. In two plots of a lightly exploited abalone population in South Australia, all the greenlip abalone (Haliotis laevigata Donovan) were enumerated by divers, providing the true density in both study regions. Clustering of abalone was visually evident and quantified by a Hopkins test. The study areas were gridded into 1-m2 quadrats. Divers measured distances from randomly selected grid points to the nearest abalone, and from that nearest abalone to its nearest neighbor. A second set of inter-abalone distances from every fifth tagged abalone were also measured. Two PNN estimator formulas, of Byth (1982) and Diggle (1975), were used to estimate abalone density. The resulting estimates from both PNN estimators were biased, underestimating true (enumerated) density by 18% to 29% and 18% to 55% in the two sites respectively. The Byth estimator showed less underestimation. Clustering of abalone is a likely cause of density underestimation in the two study areas. Simulated PNN surveys in simulated clustered populations quantified both overestimation and underestimation bias. Randomly interspersed individuals (“loners”) reduced density underestimation, and centrally (rather than uniformly) distributed clusters worsened it. Because the spatial distributions of abalone and other invertebrates are often clustered, this strong bias is problematic for the use of PNN as a survey method for estimating density in these populations.