DistanceSampling.pptJanuary 24, 2007
FW663 — Sampling & Analysis of Vertebrate Populations
FW663 is designed to include a balance of science philosophy, statistical theory and biological application. While the overall theme of the course deals with sampling and analysis theory for biological populations, the course is more broad, providing the advanced student with the following:
- Some philosophy of inductive inference (e.g., estimation of parameters and measures of precision).
- A critical attitude concerning “results” and “findings” and an appreciation of the importance of underlying assumptions.
- State-of-the-science information on sampling, analysis, and inference theory for populations in terrestrial and aquatic environments.
- Practical experience in sample design, data collection, analysis and inference in several experimental situations.
- Current publications and bibliographies useful for future reference.
- Familiarity with computer software for the sophisticated exploration of complex biological problems.
The first lectures will cover background material (to bring all students up to the same level) and general methodologies (inference methods, strong inference, and statistical concepts such as mean square errors, power of tests, likelihood functions, profile likelihoods, model selection philosophy, Akaike’s Information Criterion, components of variance, properties of estimators and information matrices).
The main section of the course will deal with estimation of population size, survival rates (finite and instantaneous), and birth rates from various types of sampling data from animal populations. The course will cover the following topics:
Sample data consisting of both marked and unmarked animals.
Capture-recapture models: Both open and closed population models. This topic covers a large class of methods allowing birth and death rates and/or population size to be estimated.
Sample data consisting of only marked animals.
- Survival estimation from tagged, banded, and radioed animals: Age and/or time-specific survival rates, constant or variable effort.
Sample data consisting of only unmarked animals.
Occupancy estimation: detection probabilities <1.
- Removal models: Constant and variable effort, closed populations.
- Catch-curve models: Constant effort, stable and stationary populations, data are recorded by age or size class.
- Generalized distance sampling: Line and point transect population sampling.
- Change in ratio models: Constant effort, closed populations, sampling before and after a class-specific population change.
The above methods often have comprehensive software available (NOREMARK plus User’s Manual; MARK including the Euring 97 paper published in 1999, Evan Cooch’s Gentle Introduction (continually updated), and papers from the 2nd IWMC in Hungary; DISTANCE plus User’s Guide, RELEASE plus the User’s Manual; and CAPTURE plus the User’s Manual) and laboratory sessions will make extensive use of these packages in the Warner College of Natural Resources PC Computer Laboratory.
Major source materials will include (electronic copies of some of these available here):
Grades will be assigned based on the following weights:
|20% Quizzes (average with lowest 2 quiz scores discarded)|
|30% Midterm I|
|20% Midterm II|
|30% Final Exam|
Class participation will be used to help decide borderline grades, i.e., a person that falls on the A/B line may receive an A if class participation was high. This person would have asked questions in class, and helped keep the discussion sections interesting. In contrast, a lack of participation might mean a B for a person on the borderline between an A and a B.
Laboratory exercises will not be graded, but examined to see how well you understood the questions and the objective of the lab. You are encouraged to work together in the laboratory. Particularly important is that students with prior computer knowledge help students without prior computer training. Not understanding a laboratory exercise means you have probably missed an important concept. Lab performance is measured through the midterms and the final exam. You will likely be asked to perform an analysis in a take-home exam similar to a laboratory exercise.
FW663 is intensive, partially because it is offered on an accelerated basis. Each 4 hour period is 1.5 hrs. lecture, 0.5 hrs. recitation and 2 hrs. of laboratory.
Students would be unwise to take more than one other course concurrent with FW663. A graduate seminar or ST512 are good choices. FW663 will take every available hour for most graduate students; do not get over-committed during Spring Semester.
“Homework” is usually estimated as # credits times 2 hrs./week. FW663 is taught at 1.5 pace, thus 5 x 2 x 1.5 = 15 hours per week should be scheduled for “homework and study” for this class. Students that are statistically challenged or computer impaired should allow additional time. With 12 hours of class time and at least 15 hours of “homework” it becomes clear that FW663 will occupy a solid 30 hours per week of a student’s time.
Expect an almost daily quiz; this keeps everyone learning and allows the instructors to gauge understanding and reinforce certain points where understanding is low. The daily quiz is an important learning tool. In addition, quizzes force everyone to “keep up.”
Laboratory sessions try to facilitate team-work. Get used to working together to solve problems and gain understanding.
ALWAYS use ONLY your student ID (not your name) on all material to be turned in (e.g., Social Security Number on quizzes, mid-term and final examinations). Using only your social security number aids in objective grading of the material.
Overview of Bayesian Estimation in MARK
|Data Dredging — Two Examples|
|Goodness-of-fit of Product Multinomial Models|
|Models Versus Full Reality|
|The Set of Candidate Models|
The following papers are to be published from a Program MARK workshop in Hungary, June, 1999.
|First Steps with Program MARK: Linear Models — Evan Cooch|
|Exploring Ecological Relationships in Survival and Estimating Rates of Population Change Using Program MARK — Alan B. Franklin|
|The Robust Design for Capture-Recapture Studies: Analysis using Program MARK — William L. Kendall|
|Jointly Analyzing Live and Dead Encounters using MARK — Richard J. Barker and Gary C. White|
|Advanced Features of Program MARK — Gary C. White, Kenneth P. Burnham, and David R. Anderson|
The following papers are published in the proceedings of Euring 1997 in Bird Study Volume 46.
|Understanding Information Criteria for Selection among Capture-Recpature or Ring Recovery Models— David R. Anderson and Kenneth P. Burnham|
|General Strategies for the Collection and Analysis of Ringing Data — David R. Anderson and Kenneth P. Burnham|
|Program MARK: Survival Estimation from Populations of Marked Animals — Gary C. White and Kenneth P. Burnham<|
Dr. Nigel G. Yoccoz, Department of Arctic Ecology, Norwegian Institute for Nature Research (NINA), Polar Environmental Centre, N-9296 Troms�, NORWAY, presented a slide show in the workshop session on the multinomial likelihood at the Euring 2000 meeting.
There was once a group of Statisticians and a group of Engineers riding together on a train to joint meetings. All the Engineers had tickets, but the Statisticians only had one ticket between them. Inquisitive by nature, the Engineers asked the Statisticians how they were going to get away with such a small sample of tickets when the conductor came through. The Statisticians said, “Easy. We have methods for dealing with that.”
Later, when the conductor came to punch tickets, all the Statisticians slipped quietly into the bathroom. When the conductor knocked on the door, the head Statistician slipped their one ticket under the door thoroughly fooling the layman conductor.
After the joint meetings were over, the Statisticians and the Engineers again found themselves on the same train. Always quick to catch on, the Engineers had purchased one ticket between them. The Statisticians (always on the cutting edge) had purchased NO tickets for the trip home. Confused, the Engineers asked the Statisticians “We understand how your methods worked when you had one ticket, but how can you possibly get away with no tickets?” “Easy,” replied the Statisticians smugly, “we have different methods for dealing with that situation.”
Later, when the conductor was in the next car, all the Engineers trotted off to the bathroom with their one ticket and all the Statisticians packed into the other bathroom. Shortly, the head Statistician crept over to where the Engineers were hiding and knocked authoritatively on the door. As they had been instructed, the Engineers slipped their one ticket under the door. The head Statistician took the Engineers’ one and only ticket and returned triumphantly to the Statistician group. Of course, the Engineers were subsequently discovered and publicly humiliated.
MORAL OF THE STORY. Do not use statistical methods unless you understand the principles behind them.