Statistical Issues for Monitoring Ecological and Natural Resources in the United States

The United States funds a number of national monitoring programs to measure the status and trends of ecological and natural resources. Each of these programs has a unique focus; the scientific objectives are different as are the sample designs. However, individuals and committees, all well aware of the cost of ecological monitoring, have called for more effective monitoring programs. The objective of this paper is to summarize existing programs' statistical designs and discuss potential alternatives for improvement in national monitoring. Can we improve the current situation by providing an overall framework for the design or analysis of data from these disparate surveys? First, the paper summarizes the objectives of these surveys, compares and contrasts their survey designs as currently implemented, and determines what variables they collect. Through this process we identify commonalities and issues that impact our ability to combine information across one or more of the surveys. Three potential alternatives are presented, leading to comprehensive monitoring in the United States.     

EMAP Design and River Reach File 3 (RF3) as a Sample Frame in the Central Valley, California

The Central Valley, California, R-EMAP project assessed the effects of highly modified, agriculturally dominated landuse on the aquatic resources of the lower portion of the Central Valley watersheds. The focus of this paper is to assess the utility of the EMAP design and the River Reach File version 3 (RF3) 1:100,000 scale Digital Line Graph (DLG) as a sampling frame. The study area is 34,099 mi²(88,316 km²) and comprises the lower reaches of the Sacramento River and San Joaquin River watersheds to the 1000 ft. (305 m) elevation. Sampling sites are selected using a tessellation stratified design to represent the two main populations of interest: natural streams and man-made waterways. Sites are selected to represent 13,226 miles of streams and sloughs, and 14,648 miles of irrigation canals, ditches and drains. To achieve an approximately equal sample size across stream orders and basins, the sample design was weighted by Strahler order categories to ensure sampling occurred in the higher order streams. Based on office and field reconnaissance, the study provided information on the quality of RF3 as a sampling frame. Site selection using RF3 had a success rate of approximately 44%. The RF3 database has an error rate of approximately 7%. When human influence factors were included, the error rate increased to 16%. There was an 11% error rate when selecting sites for natural streams, and approximately a 14% error rate for man-made waterways. The reconnaissance information indicated that presence or absence of irrigation ditches and return drains depends on changing agricultural uses. Some of the error in the RF3 for natural streams and man-made waterways can be attributed to rapid urban expansion, especially in the San Joaquin basin.