DURHAM, N.C. – It’s a paradox that has puzzled ecologists for half a century: Species in competition, such as trees fighting for precious sunlight in a forest, are supposed to co-exist well by responding differently in the battle for limited resources. But those differences never show up in the data when ecologists measure a forest.

Competitors like black gums and red maples, for instance, have coexisted for millennia in the shaded understories of eastern U.S. forests, yet species-level data offer scant proof that they respond differently to environmental fluctuations that limit access to light, soil moisture and other essential resources – the very differences required for their long-term coexistence.

Are the models and theory flawed? Or it is the data?

A study by a 91 forest ecologist in the February 19 issue of Science offers new insights that may resolve this paradox once and for all. 

“Species differences do exist, consistent with theory, but species-level data don’t show them. In order to see them, you have to go to the individual level,” says James S. Clark, H.L. Blomquist Professor of Environment at Duke’s Nicholas School of the Environment, and professor of Biology and Statistics.

Scientists working to understand a variety of pressing ecological issues, including the spread of invasive species, could benefit from this finding.

Species-level studies – the preferred approach in nearly all past research on forest tree diversity – rely on average responses from sample populations to infer average growth, reproduction and survival rates for entire species over time. They are useful for many purposes, Clark says, but because they assess species’ responses in only the handful of environmental dimensions that can be measured, they miss most of the subtle ways in which species differ.

“The environment varies in thousands of ways. Species can differ in how they respond in all of these dimensions, and we can only measure a few of them,” he says.

Failure to find tradeoffs in how species respond in the few measurable dimensions, such as the ability to survive in a shaded understory versus growing fast in full sunlight, has promoted inaccurate explanations for diversity. “Looking at only a handful of averaged responses in a few dimensions can lead to the conclusion that all species react more or less the same,” Clark says. “That’s what we’ve been doing until now – searching for ways that species differ in a few dimensions. With species-level data, you don’t see variations among individuals within populations. Yet this is what provides the evidence that species differ in their distribution of responses to the environment, despite the fact that the populations to which they belong do not differ, on average.”

To avoid this shortcoming, Clark analyzed more than 226,000 “tree years” of data from more than 22,000 individual trees, from 33 different species, in 11 forests in three regions of the Southeast.

He estimated annual rates of growth, fecundity and survival risk for every tree in the 11 stands from observations of tree diameter, canopy spread and height, reproductive status, survival, and seed dispersal. Using a hierarchical Bayesian analysis allowed Clark to quantify variations between individuals within populations over time spans ranging from six to 18 years.

The species-level analysis, from this study funded by the National Science Foundation, failed to reveal any of the subtle differences that the models say are needed to explain high diversity. However, the variation needed to promote diversity was clearly evident at the individual level.

“I found that individuals are responding in many dimensions in different ways, but with more similarity to other individuals of the same species than to individuals of different species,” Clark says.

That’s important, he explains, because “as the individuals in a population react to a variable environment, the similarity they share with others of their own species tends to concentrate competitive pressure within species. The tendency to compete within the species allows multiple species to share a high-dimensional landscape.”

These findings, he notes, are consistent with a classical result from ecological theory – that coexistence of competitors requires stronger competition within than between species.

“As long as studies focused on species-level averages, we couldn’t see how this requirement could be achieved. By moving to the individual level, we see the pattern: individuals responding more like others of the same species,” Clark says.

The change in perspective yields new insights into a variety of pressing ecological issues.

“We’ve always wondered, for instance, how introduced species could invade ecosystems where competition is already intense. The assumption was: Since existing species already are competing for limited resources, it must be especially difficult for invaders to come in, establish and compete successfully,” Clark says. “But these results suggest that competitive exclusion is less efficacious than we previously thought. Knowing that diversity is likely controlled by variations in many, many dimensions makes it easier to understand why invasions can be so common, and suggests new ways of thinking about why they can be so successful.”