In the summer of 2005, I visited for two months Hamish McCallum and Hugh Possingham's labs at the University of Queensland in Brisbane, Australia. Hamish had done some work with my PhD advisor on diseases and I thought visiting his lab was an excellent opportunity for broadening my scientific horizons. I was also looking forward to going back to Australia after a very fun vacation in the summer of 2003 following the end of my Master's.
Hamish and his wife were terrific hosts (I didn't have any interaction with them after my 2005 visit, something I regret) and I remember a nice, although brief, conversation with Hugh Possingham (Full Professor at the University of Queensland, who I had met the year before at a workshop in Trieste) about my PhD research (I met Hugh multiple times in later years, always brilliant and very gracious). Hugh is one of the most talented and relevant conservation biologists in the world (now also Chief Scientist of the Nature Conservancy) and he offered some suggestions on more conservation-motivated studies on marble trout. I still think about those suggestions, which have been a recurrent theme in my research statements since then*.
I also visited my friend Donna from Melbourne who I had met for the first time in 2003 (one of the kindest people I have ever met in my life. I have been trying forever to get back in touch with her, but her email address apparently changed and she has no presence on social media) and my former soccer teammate Gianfranco Circati, who had moved to Perth with his family the year before. Overall, I had an amazing time in Australia, personally and professionally**.
My research on spatial distribution models of species was steadily going forward. The first paper I published on species distribution models for clams is still my most-cited work (~80 citations as of March 2017. Vincenzi, S. et al. 2006 A GIS-based habitat suitability model for commercial yield estimation of Tapes philippinarum in a Mediterranean coastal lagoon (Sacca di Goro, Italy). Ecol. Modell. 193, 90–104). I refined an idea of my PhD advisor about combining data and expert knowledge to get a better understanding of the spatial distribution of clams and of the commercial yield potential of different areas within coastal lagoon. Clams can grow well in certain areas, but due to constraints such as water depth or accessibility by boat, the commercial yield potential of that area could be close to zero. In applied work, I firmly believe that expert knowledge has to be seriously taken into account, as it often can provide better or faster insights (they are already there, just ask them!) that those provided by more formal methods of investigation. Fishermen are there to make money and they do not waste time harvesting or fishing where there are no clams or fish (or mushrooms, if we extend the analogy to mountain people, but don't ask them because they are not going to tell you where the mushrooms are) or where clams or fish cannot be harvested or fished.
I presented the results of my work on clams at the 2004 annual meeting of the Italian Society of Ecology (basically 4 months after the start of my PhD) and I won the award for the best talk given by a young researcher in ecology (along with 600 euros if I remember well, handed over by the at-the-time Italian Minister of Environment Altero Matteoli). Fiorenza Micheli, now Full Professor at Stanford, was in the award committee; supported by my PhD advisor, I asked her to visit her lab in the summer of 2006. It was my first trip to the US and one of the turning points of my academic career.
Let's get back to fish. Due to the strong relationship between egg production and body size of females in salmonids, I also became interested in finding out the determinants of variation in fish body size and estimating their relative importance. I still didn't have a clear vision for my research, which was truly driven by the curiosity of the month or of the week. However, one of my long-standing interests was understanding variation in phenotypes in humans and their environmental determinants, and I was reading with a certain interest about predictive adaptive responses and the famous Dutch famine study.
Predictive adaptive responses are physiological or behavioral changes in an organism that become fixed after early stimuli. For example, it is now well known that growing up in an unstable family (drugs, alcohol, violence) increases the chances of having pregnancies (too) early in life and of being sexually more adventurous. The hypothesis is that the most parsimonious prediction on the quality of the adult environment is that it will be similar to the environment experienced early in life; since life is predicted to be short and brutish when growing up in an unstable environment, it is adaptive (that is, it increases fitness) to try to reproduce as fast as possible (see Nettle, D. et al. 2010 Early-life conditions and age at first pregnancy in British women. Proc. Biol. Sci. 278, 1721–1727). Simply, if you wait too long, you might not be there anymore, so better trying to have fun earlier than never.
My research on predictive adaptive responses had also some profound implications for my view of the world. Much of our behavior is outside our conscious control and very often the results of our genetic make-up and of things that happened when we were babies (or even before when surrounded by the placenta), but easily "rationalized" a posteriori. Years later during a science retreat, I told one of my colleagues that I was surprised that so many scientists rarely apply their scientific skills, thinking, and findings to everyday problems. He did not get my observation, which kinda proved the point. On the other and darker side of the moon, I read a blog post written by a scientist a few years ago in which she was trying to motivate herself to exercise by reviewing papers that found a positive relationship between physical exercise and health, mental sharpness etc. I was both amused and appalled, but mostly appalled. It's like reviewing papers describing how water extinguishes bonfires.
Back to fish again. I saw an opportunity to test some general hypotheses on the determination early in life (we can also call it "imprinting") of adult phenotypes using trout as a model system. In this context, the main differences between humans and fish are that fish continue to grow after sexual maturity, albeit at a much slower place, and the far bigger role of environment (food, temperature) relative to genetics in determining body size in fish with respect to humans. Marble trout (mostly marble-brown hybrid) can reach 20 kg when living in lakes or reservoirs and close to 1 kg when living at higher altitudes***.
As can be easily inferred by the steady increase of mean height in humans living in Western societies after WWII, environment (better food, medicine and more hygienic conditions limiting the effects of diseases, thus allowing the allocation of more energy to growth) plays a role in human growth, but current mean height of Italians is no more than 10% greater than their pre-WWII mean height. It felt like I was connecting my research on marble trout with something I had a deep interest for. Predictive adaptive responses and early determination of early phenotypes became a theme underlying - implicitly or explicitly - much of my research****. I found out that the mean body size at age 1 of trout born in the same year (a year-class) well predicted their mean size at later ages, and that mean body size of a year-class was well predicted by how many other individuals were living in the population when the year-class was born. Then, fish relatively smaller early in life were also relatively smaller later in life, there was little chance to catch-up. I published the results in two papers and years later I used much more sophisticated statistical methods to further refine my insights.
I also became interested in the role of extreme events in determining the risk of extinction of these small (typically a few hundred fish) and isolated (fish cannot move from one stream to another) marble trout populations. Alain Crivelli (a fantastic field biologist who started the marble trout project and recently retired, we had a great, stimulating, and productive collaboration over the years) and Dusan Jesensek (manager of the Tolmin Angling Association, solid leader of men, very bright, dedicated, and with great skills for field work) were also interested in the effects of flash floods. In 2004, flash floods and debris flows led to the extinction of one population of marble trout (living in a stream called Gorska) and a noticeable reduction in population density in some other populations. What if the main determinant of risk of extinction was the frequency and intensity of flash floods or debris flows, and survival, growth, and reproduction played a much smaller role?
The main problems were that, first, we did not have any information on water flows in the mountain streams that we were monitoring and, second, the gauges installed in the main rivers (Soca, Idrijca, Baca) could not be used to infer water flows in their tributaries, those small mountain streams in which marble trout live. Weather in Western Slovenia is highly local, and the effects of bad weather on areas only 10 km apart can be (and often are) vastly different. Although there was a bit of circular reasoning, we knew that an extreme event occurred when we observed massive mortality of fish along with visible modifications of the habitat, such as huge boulders that had moved downstream, mud everywhere, or downed trees.
I proceeded with a scenario analysis (what if the time of return of flash floods is 10 years? 20 years?) and I wrote a very nice paper with my colleagues that was later published in Biological Conservation (Vincenzi, S. et al. 2008 Potential factors controlling the population viability of newly introduced endangered marble trout populations. Biol. Conserv. 141, 198–210). Apart from hybridization with brown trout and flash floods or landslides, marble trout were doing well. And why should we have expected anything different? Generally, changes of state are rare (while playing soccer I was often saying that I could easily predict the injured of tomorrow by looking at the injured of yesterday), and marble trout had been living in those streams for centuries and maybe thousands of years. At that point, I expected population viability analysis for marble trout to be largely a waste of precious research time, and I started to get more and more interested in the evolution of traits following flash floods and in investigating which mechanisms were allowing fish to bounce back to normal abundance levels after they got heavily reduced in numbers by flash floods. While one population got extinct, others were recovering fast after getting hit by a flash flood.
* In 2004, flash floods and debris flows led to the extinction of one population of marble trout (Gorska) and a noticeable reduction in population density in other populations. Hugh suggested trying to estimate the correlated risk of extinction of multiple populations following weather extremes. In 2013, I collected a huge amount of data on daily rainfall all over Slovenia (>200 stations, records start typically in the 60s) with the goal of estimating a spatial correlation structure of rainfall extremes. Data are still there, but I never had the time to properly analyze them. It would be an amazing project for a PhD student in ecological statistics or applied math. Send me an email if you are interested.
**I still have this memory of me coding in R at the bar of the hostel in Brisbane where I was staying. Before moving to the hostel, I was staying at a cottage quite far from UQ, going there by bike was like doing the Paris-Roubaix every morning. Anyway, the R code was for a bootstrapping procedure I used for estimating survival of marble trout in the first year of life (Vincenzi, S. et al. 2007 Early survival of marble trout Salmo marmoratus: Evidence for density dependence? Ecol. Freshw. Fish 16, 116–123.). I knew very little about bootstrapping, MonteCarlo procedures and similar simulation-based methods, but it always felt intuitive to me using simulation-based approaches (also because I do not like very much to use pen and paper to find analytical solutions) for statistical inference and scenario analysis.
*** I regret not having invested more time understanding how much variation in growth in marble trout is due to variation in food. Offspring of wild trout growing in fish farm fed ad lib grow a bit faster and bigger than trout growing in the wild, but fish growing in reservoirs can become huge (up to 20 kg). Those fish living in reservoirs are marble-brown hybrid, but hybrid vigor alone cannot explain the huge variation in size. My, Dusan Jesensek, and Alain Crivelli's hypothesis to explain the variation in size between fish living in the reservoirs and fish living in mountain streams was piscivory. Although marble trout living in the wild can be cannibals (and many of them are), in reservoirs they have access to more fish and a larger variety of species (only marble trout are living in those mountain streams). They start eating fish, they become bigger, they can eat bigger fish because their mouth gets bigger and so on. However, it is difficult to understand why cannibalism cannot provide a similar positive feedback (large marble trout as far we know do not eat other marble trout that are age 4 or older) or why cannibalism does not lead the population to extinction. If they can eat one little marble, why don't they eat ten? I also found fascinating that goldfish living in small bowls become smaller than goldfish living in bigger bowls. Can "sensing the environment" explainin part why marble trout do not grow as big as they genetic potential would allow? As far as I know, there is no literature on this topic. We talked about doing isotopic analysis on trout living in the reservoirs in 2013, but there was not enough money or not enough interest. Oh well.
**** As I wrote in the Discussion of my 2014 PLoSCompBio paper: "A sizable literature on prediction of future growth exists for humans, especially in the context of early identification of pathologies [...]. An approach similar to that presented in our work for the estimation of lifetime growth trajectories given only information on growth and size during the early stages of life was proposed in [...] and [...]. In particular, in [...] Shohoji et al estimated the lifetime growth of Japanese girls using measurement up to the age of 6 years old. They first adapted to humans a parametric model previously developed to model the growth in weight of savannah baboons.". The connection of human growth was still there (I also used birds later on to investigate similar processes).