When I was in academia—before I moved into technology and began applying statistical modeling and machine learning to a variety of tech problems—my research in mathematical biology was primarily on understanding what happens to populations and life histories of individuals (i.e., the timing of key events in an organism's life and the trade-offs between life-history traits, e.g., growth, survival, movement, fecundity) that are affected by catastrophic events, such as floods, storms, and fires that cause mass mortality.

I summarized and then tested using freshwater fish populations as model systems some hypotheses about what happens to populations and individuals after catastrophic events in the paper "Genetic and life-history consequences of extreme climate events" published a few years ago in Proceedings of the Royal Society B (https://simonevincenzi.com/Publications/Vincenzi%20et%20al.%202017%20PRSB.pdf) and in some subsequent work that can be found at https://simonevincenzi.com/publications/. I combined genetics, life-history theory, field studies, and simulations, and had a lot of fun doing it. Later, I applied many of the tools, models, and ideas I developed and learned when studying extreme events to problems I faced when working in technology (https://simonevincenzi.com/Publications/biology_and_movies_v2...).

A major difficulty in generalizing the consequences of disasters is that the effects of extreme events are largely context-specific. For example, the emergence of adaptations may depend on both the life histories of species and the recurrence, intensity, and nature of extreme events. Next, the demographic and genetic effects of extremes are often the result of chance and thus are not easily predicted or generalized across species or habitats.

Luck also plays a role in determining whether a population will recover after a population crash. For example, I found (combining field studies and novel genetics tools and findings, such as how to assign sex to salmonids and how to infer trios mother-father-offspring in highly inbred populations) that the nearly complete recovery of a fish population that had been reduced to a handful of individuals after a flash flood was due to the large production of offspring from a single pair. Given the high variance in adult reproductive success in most animal species (including humans)—which is at least in part due to differences in individual "quality" (a tremendously important, but I believe understudied, trait)—if the vigorous mating pair had been killed or moved during the flood, population recovery would have become much less likely.

As Napoleon famously said: "I'd rather have a lucky general than a good one."