Sponsoring Section/Society: WNAR
Session Slot: 10:30-12:20 Thursday
Estimated Audience Size: xx-xxx
AudioVisual Request: xxx
Session Title: Application of Adaptive Cluster Sampling to
Biological Populations
Theme Session: No
Applied Session: Yes
Session Content: Public policy is linked to management of biological
populations because population changes can have differential consequences
to society. For example, society faces tradeoffs when conservation of
rare or endangered species interferes with other public and private uses
of land, water, or mineral resources. Alternatively, consumers of
commercially valuable species, such as ocean fisheries, can have their
interest curtailed when population size drops to a point where managers
restrict harvest. In either case, accurate estimates of population
parameters are critical to successful conservation and informed public
debate. However, due to spatial clustering and rarity of many biological
populations, population parameters are often difficult to estimate
accurately, and yield of individual plants or animals is often low.
Application of conventional finite-population sampling schemes to rare
and clustered populations may not provide sufficient information to make
effective policy given the sampling effort usually afforded to biological
surveys.
Adaptive cluster sampling is a class of unconventional sampling techniques specifically designed for sampling rare and clustered populations. It increases sampling effort where high abundance is observed. Its application to study of highly aggregated biological populations promises more information - more precise estimates, higher yield of individuals - compared to non-adaptive sampling techniques (e.g., waterfowl wintering in central Florida, Pacific hake larvae off California coast, freshwater mussels in eastern North American rivers). However, questions still remain about its application: what degree of rarity and clustering is required for adaptive cluster sampling to be efficient, how can design factors be set optimally, how can final sample size be limited? Answers to these questions likely will precede widespread application of adaptive sampling techniques to the study of biological populations.
This session will bring together researchers active in the theoretical
development and application of adaptive cluster sampling techniques to
present and discuss recent developments, effective application, and
important areas of research.
Session Organizer: Lo, Nancy C.H. National Marine Fisheries Service
Address: Southwest Fisheries Science Center P.O. Box 271 La Jolla, CA 92038
Phone: 619-546-7123
Fax: 619-546-5656
Email: nlo@ucsd.edu
Co-organizer of the session is David R. Smith, U.S.
Geological Survey, Biological Resources Division. He is also the third
speaker and contact information for him is provided below.
Session Timing: 110 minutes total (Sorry about format):
Opening Remarks by Chair - 5 minutes First Speaker - 25 minutes Second Speaker - 25 minutes Third Speaker - 25 minutes Discussion lead by Discussant - 20 minutes Floor Discussion lead by Chair - 10 minutes
Session Chair: Lo, Nancy C.H. National Marine Fisheries Service
Address: Southwest Fisheries Science Center P.O. Box 271 La Jolla, CA 92038
Phone: 619-546-7123
Fax: 619-546-5656
Email: nlo@ucsd.edu
1. Restricted Adaptive Cluster Sampling
Brown, Jennifer, Department of Mathematics, University of Cantebury
Address: Biomathematics Resarch Centre Department of Mathematics and Statistics University of Canterbury Private Bag 4800, Christchurch New Zealand
Phone: 03-364-2987 ext 7684
Fax: 03-364-2587
Email: J.Brown@math.canterbury.ac.nz
Abstract: Adaptive cluster sampling can be an efficient design for sampling rare and patchy populations. With this design the initial sample size is fixed but the size of the final sample (and total sampling effort) can not be predicted prior to sampling. For some populations the final sample size can be quite variable depending on the level of patchiness. Restricted adaptive cluster sampling is a proposed modification where a limit is placed on the sample size prior to sampling and quadrats are selected sequentially for the initial sample. As a result there is less variation in the final sample size and the total sampling effort can be predicted with some certainty, which is important for many ecological studies. Estimates of density are biased with the restricted design but under some circumstances the bias can be estimated by bootstrapping.
2. Sequential Sampling Designs to Estimate Abundance of Rare Populations
Christman, Mary C., American Univeristy
Address: Department of Mathematics and Statistics American University 4400 Massachusetts Ave NW Washington, DC 20016-8050
Phone: 202-885-3123
Fax: 202-885-3155
Email: xman@american.edu
Abstract: Of interest is designing sampling schemes for rare populations where we utilize information about the degree of clustering of the elements being sampled. Some existing designs, such as adaptive cluster sampling, have been developed that exploit that information but which can have some problems associated with them. These include a random sample size and a design which behaves poorly when the population is neither rare nor highly clustered. We consider a sequential sampling design that combines the positive aspects of adaptive sampling and of sequential analysis. Several sampling designs that are based on taking samples sequentially until some predefined stopping condition is met and which incorporates an adaptive sampling component are developed. For the designs considered, estimators and measures of their accuracy are given and their behavior analyzed.
3. Observations of a Practicing Adaptive Sampler: Lessons Learned, Conclusions Drawn, and Questions Remaining
Smith, David R., Biological Resources Division, U.S. Geological Survey
Address: U.S. Geological Survey Biological Resources Division 1700 Leetown Rd. Kearneysville, WV 25430
Phone: 304-725-8461x383
Fax: 304-728-6509
Email: david_r_smith@usgs.gov
Abstract: Adaptive cluster sampling methods are intuitively appealing because they permit biologists to do what comes naturally - increase sampling effort where rare organisms are discovered. However, it is my impression that while biologists are intrigued by the technique, widespread application is lagging. Through a review of the design of a biological survey that makes use of adaptive cluster sampling, I will highlight the decisions that give biologists pause. Much of the problem in design of such a survey is avoiding over and under sampling. I will illustrate potential problems and make some recommendations based on how I have practiced adaptive cluster sampling. Finally, I will question whether increased efficiency is sufficient justification for application of adaptive cluster sampling, especially for rare populations; rather I will suggest that increased yield of individuals and species is its main benefit.
Discussant: Thompson, Steve K. Pennsylvania State University
Address: Department of Statistics Pennsylvania State University 326 Classroom Bldg. University Park, PA 16802-2111
Phone: 814-865-3235
Fax: 814-863-7114
Email: skt@stat.psu.edu
List of speakers who are nonmembers: None ????