Deborah M. Gordon: Ant Encounters: Interaction Networks and Colony Behavior. (2010)
Sat, Feb 08, 2025./noise/books | LinkGordon is an eminent myrmecologist at Stanford. This book is the first in the Santa Fe Institute's Primers in Complex Systems series, and as such I was expecting more than just the usual pop-sci compilation of (fascinating) ant tales of the kind that, for instance, Hölldobler and Wilson and Moffett have recounted over the years. There are no photos, just some figures taken from research papers. There is no mathematical modelling. Genetics receives minor (but interesting) coverage.
The main focus of the book is the concept of an "interaction network" — roughly a model (familiar from social networks) of how ants communicate with each other and how that feeds into behaviour. Some experiments are described and proposed, chiefly constrained by an unwillingness to destroy ant nests. (One reason amongst many is that Gordon aimed to study colonies over decades.) The encounter rate is key! Some observations seem to miss the obvious. For example, we're told that if the rate at which first-mover patroller red harvester ants Pogonomyrmex barbatus return to the nest falls below some threshold then the foragers for the colony are inhibited from departing, but not what happens if the patrollers return more quickly. (One can imagine the latter is irrelevant, ignored or signals an avoidable catastrophe like a flood or predator.) Unlike honeybee researchers, Gordon does not appear to have a non-invasive way of determining how much food is stored in the nest, leaving me wondering how the colony's hunger is signalled and what effect that had on forager behaviour; again one could imagine sufficient hunger leads to a majority of ants foraging whatever the success of that foraging.
Most interesting to me was her refutation of W. D. Hamilton's explanation for why it is genetically beneficial for workers to raise sisters rather than their own progeny (in Chapter 6, section Evolution of Colony Organization). It's straightforward: queens typically mate with multiple males, and this means that sister-workers may share less than 50% of their genetics on average (rather than the 75% suggested by haplodiploidy with a singular father). This point is so obvious that I feel it must've occurred to Hamilton.
Despite the promise of networks we're only told about pairwise interactions between ants performing particular tasks. Much is made of their limited attention span — about ten seconds — which I guess precludes much path dependency (etc) unless they too have some kind of fast and slow neurology. There is a division of labour but the division is not static (i.e., not determined solely by caste). No connection is drawn to how cells in a multicellular organism specialise despite an early claim that this study and that are related. Another round of editing would've helped: often the explanation for a technical term or concern occurs well after its first use or is repeated in short order. Gordon seemed overly responsive to pop cultural representations of ants.
Generously summarised at length by Leon Vlieger. Data availability is the limiting factor. Hamilton is further demolished by Seirian Sumner in Endless Forms: The Secret World of Wasps (2022). Goodreads.
