The Tree Room : How to build a tree :
Confidence in connections
Even after bringing the currently available evidence to bear and applying our best tree-building methods, a single "perfect" tree may not (and often doesn't) result. Biologists can usually uncover the major connections in a group (e.g., we are quite confident that fungi are more closely related to animals than to plants and that lizards are more closely related to mammals than to amphibians), but for a few branches, several different hypotheses about their relationships may be equally well supported by the evidence. For example, based on evidence from two genes, biologists are quite confident that the American hard pine species occupy their own branch of the Tree of Life and form a clade, a group of lineages that includes all the descendants of a common ancestor and no others. However, scientists are much less certain about other details of the tree e.g., how the knobcone, bishop, and Monterey pines are related to one another.6 Three different hypotheses about their relationships (shown below) are equally well supported by the genetic evidence.
Three possible relationships of the knobcone, bishop, and Monterey pines.
A polytomy involving the knobcone, bishop and Monterey pines. These three species formed a clade in 69% of the analyses run.
Biologists represent any uncertainty about the relationships on a tree using two approaches. First, the way that branching points are drawn can indicate uncertainty about which particular species are most closely related (share the most recent common ancestor). Polytomies are nodes on evolutionary trees with more than two descendent lineages and look something like pitchforks. They communicate that the evidence doesn't strongly support one hypothesis over another regarding how the lineages stemming from the node are related and that scientists are still unsure about how these relationships should be drawn. For example, biologists represent the relationships among the knobcone, bishop, and Monterey pines with the polytomy shown below. Second, statistics can be used to indicate how confident scientists are that a particular group forms a clade. These are called branch support statistics. One of the most common statistics used for this purpose is a bootstrap value. To get bootstrap values, a biologist builds the same tree many times, each time leaving out some portion of the evidence. Then he or she studies the trees produced to see if the same clades appear in many of them. If the same clades keep coming up, even when much of the data is left out, those clades have strong support and will receive a high bootstrap value. Scientists look for bootstrap values above 95% in order to feel very confident in a clade. Clades that are only supported by a few characters will receive lower degrees of support. In the example to the right, the knobcone, bishop, and Monterey pines formed a clade 69% of the time that the evidence was sampled. That's a fairly weak level of support. New evidence might refute the hypothesis that they form a clade.