Weighted finite population sampling to maximize entropy

Attention is drawn to a method of sampling a finite populationof N units with unequal probabilities and without replacement.The method was originally proposed by Stern & Cover (1989)as a model for lotteries. The method can be characterized asmaximizing entropy given coverage probabilities _i, or equivalentlyas having the probability of a selected sample proportionalto the product of a set of ‘weights’ w_i. We showthe essential uniqueness of the w_i given the _i. We present twomethods for stepwise selection of sampling units, and correspondingschemes for removal of units that can be used in connectionwith sample rotation. Inclusion probabilities of any order canbe written explicitly in closed form. Second-order inclusionprobabilities _ij satisfy the condition O < _ij < _ij, whichguarantees Yates & Grundy‘s variance estimator tobe unbiased, definable for all samples and always nonnegativefor any sample size.     

Multi-stage sampling for population estimation

Summary Multi-stage sampling is a convenient technique suited to the desnity estimation of biological populations living in habitats with complicated structures. This paper describes a general method of its application to population estimation in which the preliminary information on the spatial distribution pattern of the population under study can be incorporated as the parameters of the mean crowding-mean relationship. The formulae that are necessary to perform sequential or double sampling plans for its efficient application are derived. The procedure of application of the method is explained with a numerical example. This study was supported by science research fund from the Ministry of Education.     

The development of sampling strategies for population studies of coral reef fishes. A case study

Fish ecologists need to do pilot studies to develop accurate, precise and efficient sampling strategies. This paper presents a case study of the pilot investigations carried out to achieve this for three species of butterflyfish Chaetodon rainfordi, C. plebeius and Chelmon rostratus, at One Tree Reef. The effects of different transect sizes and methods on density estimates and their precision for each species were assessed. Varying transect dimensions (25, 50, 75 and 100 m lengthx1,2 and 3 m width) did not significantly affect the density estimates. However for C. rainfordi the precision of estimates was variable with the smallest transect length and width producing the least precise results. Higher density estimates were obtained for juvenile fish over a 1 m strip width than a 2 m width. A significant effect of disturbance on the densities of C. rainfordi and Chelmon rostratus was caused by the laying out of the transect tape. Consequently a technique of simultaneously censusing and delimiting transect boundaries was tested and found to give consistently higher density estimates. The potential effect of censusing at different times of the tide was examined and found to be insignificant for two species but quite complex for the other. Finally, a pilot sampling program was done at seven localities across One Tree Reef to determine the most efficient way of allocating sampling effort, for future census work.     

A methodological approach for non-invasive sampling for population size estimates in wild boars (Sus scrofa)

Composite microsatellite genotypes were determined at five loci from 35 tissue-sampled wild boars and used as reference genotypes to estimate both allelic drop-out rate and false allele rate in comparison to genotypes from scats and hair strands of the same animals. These rates allow to assess the genotyping reliability when only non-invasively collected material is available. Polymerase chain reaction (PCR) amplification from scats was often corrupted by inhibitors and worked poorly, whereas genotyping success in hair samples was high. Body region of hair origin had no influence on PCR suitability, whereas the type of hair had. We recommend the use of bristles. PCR conditions were optimized for single-hair (bristle) genotyping.