Damaged Leaf Dataset

Update Feb 2022: Thank you to the College of Arts and Sciences at UVM for awarding us a Small Grant Award to photograph the leaves preserved for this dataset. Photography will begin in March 2022.

Over the summer and fall of 2021, I closely observed the rapid life cycle of the Lymantria dispar, particularly its impact on the foliage of Oak and Maple trees in Vermont. In Chittenden County, this was the most severe Lymantria dispar outbreak in 30 years.(1) The invasive species devoured so much foliage that trees were entirely stripped of leaves. Damaged foliage dropped to the ground in summer, and weeks later, trees somewhat miraculously re-foliated new ones. My observation and documentation of this event included collecting and pressing over two thousand damaged leaves that were essentially sculpted into new forms by the caterpillar’s eating habits. The forms of the damaged leaves are majestical, terrifying, and at times comical; they look like monsters, hungry mouths, winged creatures, and appear to have animated facial expressions. Some changed in both shape and color, left looking not like leaves, but flames. Damaged Maple leaves from this outbreak are in stark contrast to the perfect Maple leaf form that symbolizes an idyllic Vermont. (See figures 1- 7.) At the height of the 2021 Lymantria dispar outbreak and before most of the caterpillars spun their cocoons, I collected the fallen and damaged green leaves. Long after the moths born from cocoons swarmed, mated, died and gruesomely disintegrated, leaving behind clusters of fuzzy yellow egg-smears clinging to the barks of trees and the side of our house, I collected again. This time, more damaged leaves fell as part of the natural autumn cycle when deciduous trees typically shed leaves with autumn colors.

The damaged leaves caught my attention in part because of my work with image generation as the director of the UVM Art + A.I. Research Group. We have been working with Generative Adversarial Networks (GANS) on the Vermont Advanced Computing Core (VACC) since the spring of 2020. Through our production of thousands of abstract image forms and then sharing them with the public, we observed the many ways humans look for familiar forms within abstract forms, somewhat like how a Rorschach test leads people to see images in arbitrary ink patterns. Additionally, the life cycle of Lymantria dispar from-egg-to-larva-to-caterpillar-to-cocoon-to-moth has a striking resemblance to working with image GANS, where vector “seeds” are cocooned within a neural network and then output as entirely new and surprising forms. The decisions made by neural networks are not transparent; (2) decision processes are essentially “black-boxed” and obscured from view, much like the transformation that takes place in a cocoon when a caterpillar transforms into a winged creature.

The aesthetic of the damaged leaves is striking and luminous. Once digitized as photographs, they become a source material that engages in conversations relevant to contemporary and digital arts, the sciences, and the humanities. More specifically, as a dataset and visual series, the damaged leaves illustrate topics such as climate change; invasive species; the sometimes-indistinguishable acts of creation and destruction; the presence and absence of ruins; and problematic binaries such as natures and cultures, natural and artificial, alien and native, and physical and digital. The common name for the insect is Gypsy Moth, a name considered an ethnic slur to the Romani people.