Adaptation between similar streams exceeds the stream-to-lake transition
A University of Bern team conducted field experiments moving sticklebacks between four seemingly similar streams and one lake, finding that stream-specific adaptations contributed 67% of the fitness advantage—exceeding the contribution from differences between habitat types themselves.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Evolution rewards identical solutions through potentially different mechanisms
The study, recently published in PNAS, comes from the Institute of Ecology and Evolution at the University of Bern. The corresponding author, Katie Peichel, previously conducted behavioral genetics research on sticklebacks at the Fred Hutch Cancer Center in Seattle before relocating her lab to Switzerland and shifting toward more field-based evolutionary ecology. The paper's title is "Dissecting unique, locus-specific, and shared habitat adaptations in a classic repeated-evolution system." Its core question: when the same species independently evolves similar traits in multiple similar environments, is this the result of adaptive selection, or just random drift coincidentally producing similar-looking outcomes?
— NelsBetween-stream variation rivals stream-lake differences, yet standard experiments miss it
The standard approach in ecology for studying these questions is reciprocal transplant experiments: move fish from lakes into stream enclosures, then move stream fish into lake enclosures, and see who adapts better. But the authors point out this only tells you that lakes and streams differ substantially; it can't answer whether "seemingly similar streams" actually harbor important, overlooked differences. So they added a third comparison: moving fish from one stream into enclosures in a different stream, where they compete against the local native population. This allowed them to separate "between-stream differences" from "stream-versus-lake differences."
— NelsRigorous field work: 715 fish in 55 cages, seven weeks of observation
The experimental setup: one lake population (Lake Constance) plus four neighboring stream populations, with enclosures at each transplant site. In total: 715 fish, 55 cages. After seven weeks, the team collected the fish and measured two fitness metrics—survival rate and body condition. Survival rate is the most direct but also strictest fitness measure: you have to live to count. Body condition is a supplementary metric: even if a transplanted fish survives, has it lost weight, potentially impairing future reproduction?
— NelsLake habitats fundamentally favor stickleback survival over stream habitats
Overall: whether native or transplanted, fish survival rates were substantially higher in the lake than in streams. Streams are the harsher habitat. This is not surprising—it aligns with the general understanding that lake resources are more stable while streams face less predictable flow and predation pressures.
— NelsAdaptation cost between similar streams exceeds the lake-stream transition
What was surprising appears in Figure 5: between-stream fitness gaps are roughly twice as large as stream-versus-lake gaps. In other words, transplanting a fish from one stream to a "similar-looking" stream costs more in fitness than transplanting it to an entirely different lake environment. Breaking this down further, the authors found that local populations' total fitness advantage over the transplanted lake population breaks down as 33% from adaptation to selection pressures shared across all streams, and 67% from adaptation to conditions unique to each individual stream—the latter is the bigger contributor.
— Nels/VincentPhenotypic convergence may mask genetic and developmental divergence
The implication: when biologists view from 10,000 feet and see several stream populations evolve similar morphological traits, they easily classify it as "parallel evolution" or "convergent pathways." But this experiment shows that identical-looking traits may actually arise from different environmental pressures and adaptive pathways. For biomedical scientists accustomed to reductionist thinking—"this trait maps to this gene"—this is a cautionary tale: you might be treating multiple distinct adaptive pathways as a single genetic mechanism.
— NelsHuman generation times prevent us from repeating stickleback-scale evolution experiments
Sticklebacks breed every 1-2 years. These four stream populations have been independently evolving since the last ice age ended, roughly 12,000 years ago—which works out to about 6,000 human generations. With human generation times of 20-30 years, observing equivalent locus-specific adaptation would require vastly longer timescales plus geographic isolation: different continents, different mountain ranges. Humans do show some regional adaptations—high-altitude Andean adaptations, lactose tolerance—but these took thousands of years plus population isolation, and modern mobility makes such rapid localized adaptation unlikely to recur.
— Vincent/NelsCCR5-Δ32 conferred advantage predates HIV by nearly two millennia
A human parallel case is the Delta-CCR5 mutation, which confers HIV-1 resistance to carriers. Yet this mutation dates to roughly 2,000 years ago, far older than HIV's emergence (around 100 years ago, early 1920s), and is carried by ~16% of the population, mostly in Europe. The true selection pressure driving it remains unclear. The mutation later played an unexpected role in the ‘Berlin patient’ case: a leukemia patient who received bone marrow from a CCR5-mutation carrier. Not only did the leukemia remit, but HIV was functionally cured as well—the virus had no cells left to infect.
— Vincent/NelsIn their own words · checked verbatim
what was really interesting here or surprising, I should say, is that the fitness difference between stream to stream was like 2x relative to – from stream to stream relative to stream to lake
Nels32:24
the fitness consequence of moving from stream to stream is actually much more sort of noted or surprisingly came out of their data
Nels33:29
of the total fitness advantage of local stream fish versus foreign lake fish, 33% reflected adaptation to selection pressures among streams
Vincent34:39
these almost like kind of creepy translucent, transparent or kind of opaque actually white worms that are various species of worms
Nels39:13
the predictability of evolution depends on the ecological resolution at which environments are defined and measured. At coarse scales, evolutionary change may often appear repeatable, whereas at finer scales, natural selection can generate a mosaic of distinct adaptive responses.
Nels45:21
these four streams in this paper, they diverged only since the last glaciation, which is about 12,000 years ago or 6,000 human generations
Vincent48:44
this thing is about 2,000 years old, so it predates HIV. And we don't know what the selection one is. It's only in certain, it's about 16% of the human population, mostly Europe.
Vincent58:02
Figures
| Experimental scale | 715 fish, 55 cages | 28:12 |
| Experiment duration | seven weeks | 28:12 |
| Between-stream fitness differences relative to stream-lake differences | ~2x | 32:24 |
| Total fitness advantage from common stream-wide selection pressure | 33% | 34:39 |
| Total fitness advantage from stream-specific local conditions | 67% | 34:39 |
| Stickleback generation time | 1-2 years | 48:44 |
| Human generation time | 20-30 years | 48:44 |
| Stream population divergence | ~12,000 years ago (~6,000 human generations) | 48:44 |
| CCR5-Δ32 mutation age | ~2,000 years | 58:02 |
| Time since HIV emergence | ~100 years (early 1920s) | 59:03 |
Glossary
- reciprocal transplant experiment
- Moving individuals from two populations into each other's habitats to measure and compare bidirectional changes in fitness.
- parallel evolution (repeated evolution)
- Independent evolution of similar traits in different populations, potentially driven by adaptation or arising by chance.
- behavioral fever
- Behavior in cold-blooded animals where they actively seek warmer environments after infection to combat pathogens through temperature elevation.
- CCR5-Δ32
- Genetic mutation causing the absence of a receptor on cell surfaces, making it difficult for HIV to invade immune cells.
How to listen
Researchers and science enthusiasts interested in field experiments in evolutionary ecology, stickleback model systems, or questions about whether convergent evolution truly converges.
First ~8 minutes of personal chat and conversation logs, and the closing segment of music and news picks.