Friday, March 11, 2022

The art of the pivot

The art of the pivot

or, when abandoning a beloved experiment leads somewhere better

Guest post by Amanda Hund


Perseverance is central to being a good scientist. We are rewarded for sticking with it, figuring it out, and making it work. On a personal level, perseverance is required for getting into and through graduate school, for learning new things, and to move forward whether from a failed experiment or a grant rejection. On a larger scale, perseverance is central to science itself, which is about exploring the unknown, asking questions that don’t yet have answers, and doing things that have never been done. Failure is a key part of the scientific process and perseverance is required for forward progress. 

            Failing and trying again was certainly part of my graduate experience. For one project, I wanted to manipulate mites in the nests of barn swallows. In the past, studies had largely relied on insecticides to do experiments with nest parasites. However, this meant that experimental designs were inherently unbalanced, as comparisons were made between nestlings exposed to parasites and nestlings exposed to chemicals. As an eager young (and definitely naïve) graduate student, I was determined to do better. My first thought was to microwave nests, something that had been done with nest box species. After some trial and error, I figured out how to use a flat blade to carefully pry off the mud swallow nests without breaking them, how to reattach them with angled nails and wet clay, how to hook a microwave up to a car battery and transport it to the field, only to discover that when you microwave a swallow nest, it usually explodes dramatically. 

Back to the drawing board. Quick acting insecticides? Nothing was quick enough. Diatomaceous earth to kill mites in the control treatment? Still an unbalanced design. I briefly toyed with the idea of building a microwave gun to disinfect nests without detaching them, only to have my partner, who happens to be physicist, politely point out that this idea was insane. Replacing nests with ones that I had frozen? The mites came back to life. Dipping nests in liquid nitrogen? They crumbled to pieces. After weeks of failing and trying again, I finally landed on using an industrial heat gun - an idea that arose from reading articles about killing bed bugs in dorm rooms and talking to my dad. It worked brilliantly. It turns out that there is a good enough buffer between the temperature needed to reliably kill mites and temperatures that cause nest material to spontaneously ignite. Using the heat gun, I was able to disinfect all the nests and add mites back into half, a balanced design! (You can read all about it here). Using this technique, I did a series of experiments looking at how mites influenced parental care, incubation, nestling immune systems, and where males establish territories. It was a clear lesson to me that sticking with it was worth it. 

Fast forward a few years and I am an eager new (and still naïve) postdoc jumping into two new systems- butterflies and stickleback fish (with advisors Emilie Snell-Rood and Dan Bolnick). For my first project, I worked with Dan to come up with an ambitious experiment involving coinfection in different stickleback populations using live laboratory infections of the tapeworm Schistocephalus solidus. Because Dan’s group had done laboratory infections with tapeworms before, most of the protocols I would need were already established. I was confident it was going to work. How hard could it be? It’s not like I needed to come up with something entirely new like the heat gun. Previous postdocs in Dan’s lab had done nearly 1000 experimental infections in prior years, so Dan agreed- projecting total confidence- and we began to plan. 
            The project started off strong. I spent several weeks in the field on Vancouver Island making crosses from wild fish to send back to the lab and collecting infected fish to breed tapeworms in the lab and get the eggs I would need for the experiment (with help from Natalie Steinel, Jesse Weber, and Jacqueline Salguero). It was around this time that Dan moved his lab from Texas to the University of Connecticut. Moving a large research lab across the country is no easy task. Everything must be taken down and set up anew. Still, I was not too worried- I figured things could just be set up in a new zip code and science would continue as planned. With my baby fish settling into the animal facility in Connecticut and my tapeworm eggs in the fridge in Texas awaiting transport, I started to finalize the details of my upcoming experiment.

The author of this post, Amanda Hund, crossing stickleback

Undergrad assistant Jacqueline Salguero wearing magnifying glasses to help sort stickleback eggs.

With most of Dan’s lab in Connecticut, it took several days for someone to notice that the lab fridge in Texas was on the fritz. By the time they found it, everything inside was hot and my tapeworm eggs had been cooked. A low point, but nothing we can’t solve! Graduate student Will Shim and I took a last-minute trip back to Vancouver Island to collect more infected fish to breed more tapeworms. We soon realized that while stickleback are very easy to catch in minnow traps in the spring while they are breeding, minnow traps are essentially useless in the fall. After several days of catching nothing, we switched to dip netting as a last-ditch effort and managed to catch a few fish to bring home, crossing our fingers that enough were infected with tapeworms. Those fish had been hard won, so we were pretty upset when we went to pack them into coolers to fly home and found them partially frozen. Unusually cold fall weather combined with the not-so-great sample fridge on the porch of the field cabin had made for less-than-ideal storage conditions. Were the tapeworms inside the fish (if there were any) now dead? I am now unable to fully trust any refrigerators. 

We got lucky. There were just enough tapeworms, and they were still alive when the fish arrived at the lab. The crosses were successful, there were enough eggs, and the experiment had been saved! Persistence, it seemed, had paid off. It was with great optimism that I traveled to Connecticut to begin the infection experiment that spring. When the copepod colony (the intermediate host needed for infections) crashed, I took it in stride. We had emergency plankton samples sent from Vancouver Island and hard-working lab techs Mariah Kenney, Meghan Maciejewski, Joseph Marini and I spent hours picking out our copepod species of choice. When the paramecium colony crashed (the copepod’s food source), I drove to Massachusetts to pick some up from the lab of Dan’s former postdoc Natalie Steinel, and quickly became an expert in both paramecium and copepod husbandry. What had been pretty low maintenance copepod care, became a daily ritual with precise and detailed protocols and careful data collection to monitor population growth. Things were not going to fail again. When the water pump for the fish tanks broke while Dan was out of town, I disassembled it, drove to an outdoor landscaping store that specialized in coi ponds to buy a new one and installed it. There was nothing I could not solve! Until the tapeworm eggs wouldn’t hatch. 

Perhaps I will write a paper one day about all the different things I tried to get those eggs to hatch (null result projects should still be published, right?). The tapeworm eggs in Texas had hatched reliably following a fairly simple protocol. I tried to replicate those conditions precisely, I changed light bulbs, borrowed a different incubator, monitored humidity and temperature in both light and dark cycles, covered and uncovered the plates with eggs at different intervals, tried different water sources. I then switched to trying published protocols from other labs. I had plates at different temperatures, different light cycles, on the counter, in the incubator, I tried plant grow lights, I tried talking to the eggs. Nothing worked. When I started testing older batches of eggs in the fridge- I found that they did hatch, under a variety of conditions. It became clear that it was likely my particular eggs that were the issue. At this point, I had come so far that I was ready to double down. Perhaps I should fly back to Vancouver and collect a third round of tapeworms from my populations of interest? My lab-reared fish were the right age and I had planned on staying in Connecticut for four months to do this phase of the experiment. Time was running out.

It was at this rather low point that Dan called a meeting. When I joined him in his office he simply said, “I think it is time to pivot.” At first, I was surprised by these words. Surely, I should just work harder to find a solution? Stick with it? But Dan was right. There are times in science when it is helpful to remember the economics concept of sunk costs. We had spent so much time and resources trying to make the infection experiment work, time and resources that we were never going to recover. But that effort should not necessarily dictate the best path forward. It certainly was not easy for me to let go of the experiment I had so carefully planned for months and months, but it had become necessary. We went back to the blank white board to start over and think about what we could do with what we had. This meeting soon spilled into a larger lab meeting, pulling in other postdocs and graduate students (including postdocs Stephen De Lisle and Foen Peng). Another postdoc, Lauren Feuss, jumped on board to help carry out the new project and it became a team effort. We brainstormed about questions and ideas, techniques we could use, how many fish we had, what samples were in the freezer, what hints and clues we could follow from other preliminary data and small pilot projects. While constrained by what was available and immediately doable, the discussion was collaborative and creative.

What emerged was the injection experiment. A simple, but dare I say elegant, experiment comparing stickleback populations to understand the evolution of parasite resistance. Using extra-fine syringes, we injected four different immune challenges directly into the fishes’ peritoneal cavities (where tapeworms grow), including: a saline control, tapeworm protein (made by grinding up frozen tapeworms), alum (a generalized immune stimulant, or adjuvant, widely used in vaccines), and a mix of tapeworm and alum. We then tracked the fish fibrosis response through time (day 1, 10, 32, and 90 time points). Videos of the fibrosis that results can be seen here. The results were clear and interesting.  Indeed, ecology and evolution one rarely gets results so clear-cut that doing statistical analyses really feels like an afterthought, the trends are so striking. All populations were able to generate a resistant response (fibrosis) to the general immune stimulant (alum), indicating that the cellular programs for fibrosis are shared and predate population divergence (indeed, another postdoc subsequently used our injection protocol to show that fibrotic response to alum is deeply conserved across the diversity of fish Vrtilek 2021). Despite this shared capacity for fibrosis, only the resistant population was able to recognize and initiate a response to the tapeworm protein alone. The resistant population also differed in the timing and resolution of their response. Our results suggested that early immune regulatory steps were the targets of rapid local evolution to the parasite. After these initial steps in the host response, selection in the resistant population likely favored rapid initiation of fibrosis, to control still-small tapeworms, and quicker resolution, to mitigate the long-term costs of fibrotic pathology. We recently published this work in Evolution Letters.

The injection experiment opened up new directions for the Bolnick lab. Several follow-up studies have already been published and more are in the works. Grant proposals are being written using what we have learned. Was the injection experiment more interesting than my original coinfection project? I can’t say for certain, but probably. It was certainly more achievable, and it facilitated collaborations that are continuing today (Lauren and I are now working together on several projects). I have not fully given up on the coinfection experiment, just shelved it for now. Maybe I will come back to it someday (when I have lots of time to troubleshoot). The experience made it clear that failure and constraints can lead to innovation and that working as a team is often more effective and can lead to more creative ideas. Both concepts that are not new in science. At a personal level, it taught me that knowing when to change directions is just as important as knowing when to stick with it. Sometimes, roadblocks to your research can be a transiently frustrating catalyst for whole new research directions, if only you are willing to let go of Plan A. 

The sampling team at Roselle Lake, BC, in 2018. Left to right: Will Shim, Amanda Hund, Jesse Weber, Jacqueline Salguero, Natalie Steinel. Photo by D. Bolnick

Roselle Lake, British Columbia

Not by vaults and locks: To archive code, or not?

  The cases for and against archiving code by, Bob Montgomerie (Queen’s University) Data Editor—The American Naturalist As of the beginn...