Musclewood flights

It has been almost 10 years since "Phase II" of Grinnell's Noyce Science Center was declared complete. An interior courtyard, planted with shade-tolerant Iowa natives (or near-natives), didn't get the memo. In forthcoming posts, some of my Ecology students will preview their studies of colonization of this habitat island by "mainland" tree species not intended to be there. Meanwhile, I'll note an independent study by my Sex Life of Plants class, in which students considered the wind dispersal capacity of a species put there on purpose (American hornbeam or musclewood, Carpinus caroliniana). Their habitat surrounded by three stories of brick and glass, these trees have little chance to colonize the rest of campus. They can, however, let students explore simple ideas about biomechanics and statistics.

A sub-sample of diaspores (fruits plus attached bracts) displays some of the range of size and shape (here, from a single tree).

When it detaches, a musclewood fruit stays aloft--and, presumably, disperses farther than it otherwise would--with the assistance of a three-lobed bract that spins, helicopter-like, as it falls. After reading an article that used multiple regression to explain and predict wind dispersal capacity (Gravuer et al. 2003), and after I persuaded them to keep their number of predictor variables few in number, easy in measurement, and plausibly simple in interpretation, the class chose to score a sample of 72 diaspores (9 for each of 8 student groups) for mass and for the length of the central-bract lobe. Dropping each three times from a height of 1.87 m (i.e., lab-bench height, plus a meter stick), they averaged drop times and regressed them on the morphological variables. They expected drop time (dispersal capacity index) to decline with mass and to increase with bract size, implicitly reasoning from the physics of terminal velocity.

Lesson 1: They were right. The regression (1.87 m drop time (s) = 0.915 - 0.03771 mass (mg) + 0.0727 bract length (mm)) and each of its coefficients are highly significant, explaining 22% of the variation in drop time.

Lesson 2: R-squared more than doubles if you add "student group" as a term.