It's not an experiment you'd want in your kitchen, but 10,000 flour beetles raised in milk bottles are changing the way scientists count animals.

In a new study to be published Friday (Oct. 19) in the journal Science, researchers used six years of beetle population data to improve the modeling tools widely used to explain fluctuations in animal numbers. That should help the people who rely on the tools for understanding or managing a variety of animal populations, including wildlife, commercial fisheries and agricultural pests.

"A primary goal of ecology is understanding population fluctuations. Our study continues that effort by teasing out more of the underlying mechanisms that drive population patterns," said mathematician Shandelle Henson of Andrews University, the study's lead author. "We want our laboratory studies to lead to useful, working concepts that can be applied to real-world problems ranging from food production to the conservation of species diversity."

The new report was written by the "Beetle Team," composed of leaders in the study of nonlinear population dynamics from several U.S. universities, and a collaborator from the Department of Environmental Science and Policy at the University of California, Davis.

This is the newest in a series of reports on population dynamics that the team has developed over the past decade using flour beetles as its animal model. In the University of Rhode Island laboratory of biologist R.F. Costantino, the rice-sized beetles (Tribolium castaneum) live in old-fashioned, half-pint milk bottles in one-quarter cup of, yes, all-purpose white flour and a pinch of dried brewer's yeast.

In the current report, the researchers compare the accuracy of two kinds of population models: the usual kind that allows animal numbers to be fractional (as in, the typical family has 2.5 offspring) and a related model that requires animals to come in whole numbers. Using each of the models, the team predicted theoretically what the beetle populations would do. Then they compared those predictions to the actual population patterns of 150 successive beetle generations.

When they compared the model predictions to the data, they found a surprise, Henson said: Neither model completely explained all the temporal patterns in population numbers. But if the researchers used both models together and also factored in random effects in birth and death events, the predictions matched the data.

The effects weave together in a mathematical space called a lattice. "Such lattice effects could be an important component of natural population fluctuations. Our study shows that a complete understanding of some population systems requires a blend of both kinds of models," Costantino said. "Now we need to get the word to the people who use these models."

The new paper, titled "Lattice Effects Observed in Chaotic Dynamics of Experimental Populations," was written by Henson and fellow mathematicians Jim Cushing of the University of Arizona and Aaron King of the University of California, Davis; statistician Brian Dennis of the University of Idaho; and Costantino and fellow biologist Robert Desharnais of California State University, Los Angeles.

UC Davis' Aaron King studies why some animal populations, such as those of the snowshoe hare and its predators, undergo cycles of extreme highs and lows. He said the new findings should be of interest to ecologists studying habitat loss, as in equatorial rain forests.

"We found that changes in habitat size can change the pattern of population fluctuations in dramatic and non-intuitive ways," King said. "We had known that, on average over time, if habitat gets smaller, a population will get smaller. But now we see that, because of lattice effects, making the habitat smaller, or even larger, can produce unexpected large fluctuations that can bring the population very close to zero -- that is, to extinction."