ABSTRACT
Spatial data consist of two components, a spatial component and an attribute component. In the sampling plans, the locations at which one collects data are generated according to a rule. In many ecological applications, the data collected by sampling at point locations on the site may be accurately modeled as varying continuously in space. Sampling error is an unavoidable consequence of the fact that the data represent a sample and not a census. There are two primary problems with random sampling of a spatial region. The first is that it is time consuming to plot out a path and travel to the randomly selected points, and the second problem is that large portions of the region may happen to go unsampled, and some regions may be oversampled. The advantage of stratified sampling of autocorrelated data is that the sample does not miss any large geographic regions.
