Model Construction and Evaluation of Line Transect Distance Data for Density Estimation

Abstract

Line transect distance sampling for estimating density was simulated in two projects using match sticks an objects of observation. Possible detection functions were constructed and constructed and modeled. The selected model in each case was used to estimate the density and compared with the true density. For line transects distance Sampling on match sticks MSDP,, the best model selected was the uniform key function plus one cosine series. It yielded a density estimate of 8.352 matchstick per square meter over the true density of 8.4. For line transect distance sampling on match sticks MSDP2, the best model selected to estimate the density was the hazard rate key. It gave an estimate of 8.12 matchstick per square meter over the true density of 8.4. In both cases selection of the chosen model was based on lowest Akaike's Information Criterion (AIC). The experience from the simulations were used to estimate the density of coconut trees in a 24 hectare farmland of Liloy, Zamboanga del Norte and the density of mangrove trees in Langamon, Anibongan, Gigaquit, Surigao del Norte. The best model selected for the Coconut Distance Project (CDP) was the uniform key function+ one simple. or Hermite polynomial, which gave a density estimate of 28.98 coconuts per hectare over the true density of 29.42. The hazard function without adjustment terms was the best model selected for the Mangrove Distance Project (MDP). It yielded an estimate of 116.52 mangroves per hectare.Results of the study, using the software package DISTANCE 3.5, showed the goodness of the estimate by distance sampling compared to the conventional strip transect full census methodology. For future research directions, a similar study might be conducted where objects of interest are mobile or clustered.