New approaches to statistics in ecology

Ecologists long ago adopted quantitative methods, and now almost every paper in major journals includes a slew of measurements. For 50 years, ecology has been dominated by a single approach to statistical inference - hypothesis testing - which became the criterion for 'sound science'. But hypothesis testing is not without its problems, notably that it engenders a false sense of certainty, running directly counter to the 'precautionary principle' which should motivate environmental management decisions.

Alternative approaches are increasingly being integrated into mainstream ecology. Although dubbed "new statistics", many of these are old ideas which could not be implemented without modern computers because of the calculations involved. More...

Course materials

These materials are regularly updated, so previous course participants may not find exactly the same materials that they received during the course. Topics without hyperlinks are still being worked on.

indicates materials used at our most recent course (Kuala Selangor, 18-21 February 2008)

R software
We recommend R and use it in most of our courses.

• Getting started with R

Basic concepts
A review of key concepts underlying traditional as well as 'new' statistics

• "Lies, damned lies, ..." - what is statistical inference?
• Squirrels - basic concepts of statistical inference
• The case of the four orang utan - probability in practice
• Frogs in ponds - maximum likelihood estimators
• Summary of probability distributions
• Issues in sampling

Models and model selection
This is a key topic and underlies most approaches to detection probability

• More frogs in ponds - AIC and likelihood
• Modeling kill rates for wolves in Isle Royale

Estimating occupancy
Simple 'presence/absence' data can give useful results, if corrected for detection probability

• Occupancy concepts
• Golden cats - data analysis with PRESENCE ( for PRESENCE v.2.1)
• Golden cats - data analysis with BUGS
• Ants - relationship between abundance and detectability

You may also like to look at the exercises on occupancy at U of Vermont.

Estimating animal densities from transects
'Distance sampling' using line or point transects give estimates of density, and hence numbers of animals in the area of interest.

• Distance sampling concepts
• Elephant dung density - data analysis with DISTANCE
• Orang utan nests - data analysis with DISTANCE

Using dung or nests to estimate animal densities
The density of animal signs, such as dung piles or nests, can be used, provided rates of production and decay can be estimated.

• Using signs to estimate densities

Estimating animal populations with mark-recapture methods
Mark-recapture methods provide estimates of the number of animals in the area sampled. If that is known, the density can be calculated.

• Rats - an exercise with a simulated rat population
• Tigers - analyzing mark-recapture data with MARK

Bayesian approaches
These methods quantify the uncertainty about estimates and produce information which can be fed into decision-making processes for management.

• Bayes in brief (includes the Cook Inlet beluga whales example)

Species richness and diversity
Biodiversity must be protected - even promoted - but is fiendishly difficult to quantify.

• The meaning of biodiversity
• Species richness - its advantages, drawbacks and uses
• Making sense of diversity indices
• A menagerie of biodiversity indices
• Bootstrapping to estimate bias and confidence intervals for diversity indices

Miscellaneous
Various topics which don't fit well under other headings

• Argus pheasant habitat
• Logging intensity and woodpeckers

References

The handouts and "lab guides" on the site are designed to be 'stand alone' documents, useful to those staff who could not attend the seminar - or anyone else who surfs to the site for that matter - as well as participants. If you find the downloads aren't sufficiently self-explanatory, or if you have other suggestions and comments, please let me know: .