| Abstract |
The statistical properties of the six measures of niche overlap of two fish species namely Leiognathus splendes and Lethrinus insulindicus using Jackknife and Bootstrap Techniques were investigated. Jackknife and Bootstrap Techniques are resampling methods used to estimate statistical properties. The six measures of niche overlap evaluated in this study are MacArthur and Levin's, Pianka's, Percentage Similarity Overlap, Horn's, Simplified Morisita's and Hurlbert's index. The statistical criteria used to describe the behavior of theses indices are the mean, standard deviation, variance, bias and confidence interval. This exploratory study was made possible by developing a computer program created the Borland Delphi Random Application Development (RAD) Tool Version 7.0. Results showed that the jackknife confidence limits do contain the observed index for the MacArthur and Levin's, Pianka's Index and Simplified Morisita's Index. Thus, these showed a successful jackknife performance for the index. However, the jackknife confidence limits do not exhibit satisfactory results. Low bias (0.0008 to 0.0386) was observed in the bootstrap performance for all the indices which revealed reliable and consistent results. Comparison of these resampling techniques showed low variance as observed in the bootstrapped performance of Simplified Morisita, MacArthur and Levin's, Pianka's and Horn's. These bootstrapped estimates are thus more efficient than jackknife. Bias estimation was observed to be lowest in the bootstrapped performance of MacArthur and Levin's indices. Jackknife performance of the MacArthur and Levin's, Pianka's and Simplified Morisita's Index also showed low bias. However, these were the lowest bias observed, bootstrapped performance was still found to be consistent in its bias estimation for all 6 measures of niche overlap. |
Mindanao State
University - Iligan Institute of Technology