Lepidoptera methodology

Here in The Big List and on the back of most of my species info sheets at the bottom, you’ll find a section titled something like “Potential caterpillar host for”. This is the product of an extremely intensive data project that I’ve undertaken to cross-reference known relationships between native flora and the larvae of native Lepidoptera (butterflies and other moths). I do this instead of sleeping.

The first phase of this project is the list of species you currently see. Reliant primarily on information shared through “Life Histories of Cascadia Butterflies” (James & Nunnallee) and the broad collaborative efforts represented at pnwmoths.biol.wwu.edu, I tracked roughly 2500 known relationships between 850 leps and 300 plant classifications ranging from species to family level.

Since the actual reality of these ecological interactions vastly exceeds what has been documented to date, I drew upon my intimate botanical knowledge to surmise what other plants would be a likely host for a given caterpillar. This is largely based on close evolutionary relationships between the plants; if a caterpillar feeds on one alder, for the most part it’s reasonable to assume that they would probably be happy enough feeding on a different alder.

More granular judgments are often needed, though. What does “aster” mean in the entomological record? There are no plants in Washington that are still classified in the genus Aster, though many were in the past. Did the researcher mean specifically that genus? Did they use it as a short-hand for the whole family of Asteraceae? Were they using it as a short-hand for herbaceous members of the family that a casual observer would call “asters”? In this case, what all Washington-native plants that have previously been classified in the genus Aster have in common is that they are members of the confusingly-named tribe Astereae, so caterpillars that have been documented to feed on “asters” are included as potentially hosted by any plant species within tribe Astereae.

This isn’t perfect, though. Within any plant group there can be considerable diversity. In the genus Artemesia, for example, you’ll find groups including woody shrubs (giant sagebrush), strongly fragrant leafy species (Suksdorf’s wormwood), and leafy species that can be relatively odorless (tarragon). I noticed that there are strong correlations between these traits and the subgenus that each species belongs to, so I extrapolated that a caterpillar that can feed on A. dracunculus could likely feed on A. campestris (subgenus Dracunculus, leafly and low-odor), a caterpillar that can feed on A. ludoviciana could likely feed on A. suksdorfii (subgenus Artemisia, leafy and odorous), and a caterpillar that can feed on A. furcata could likely feed on A. tridentata (subgenus Tridentatae, woody and odorous).

While I’m very proud of what I’ve accomplished in this first phase, I couldn’t help but dig deeper. I broadened the scope of my data collection to include the massive but much less specific database shared in the paper “Lepidoptera of North America, north of Mexico: an annotated list containing geographic ranges and host-plant records” (Shropshire & Tallamy), which is the basis for the fantastic recent publication “Moths of Western North America” (Leckie). I also re-compiled the data from the previous sources so that all caterpillar-plant relationships are ranked ranging from preferred hosts to known hosts, broad generalizations, and speculated hosts. All of the hosts are then documented to the finest taxonomic detail available, and categorized in as much detail as reasonably possible by traits that would likely be relevant to caterpillars: stature, deciduous/evergreen, wetland status, woody/leafy, clumping or rhizomatous, etc. As I move into the data analysis phase of this second iteration of The Lep List, I am developing algorithms to automatically weigh the likelihood of a given plant being the host for a given caterpillar, and then exhaustively applying my botanical knowledge to fine tune the algorithms until its predictions are ecologically plausible. The algorithms will weigh the taxonomic relationships between plant species, the plants’ traits, and how strongly a caterpillar is known to have an affinity for a particular plant.

Further, the second iteration will more methodically omit caterpillars that are known to be broad generalists that are happy to feed on most anything. They’re easy to please and of no particular conservation concern, so it doesn’t really matter what you plant for them. This first iteration of the list is generally composed of caterpillar species that are specialized enough to significantly benefit from mindfully-informed planting decisions, and the second iteration will be even more systematically focused. I have not yet decided exactly what values I will enter the algorithms to this effect, but the second iteration will probably be specifically limited to caterpillars that are known to feed either on no more than five families or no more than ten genera.