Xsciens

WorldMAP NOW

WorldMAP NOW

Tag: PNAS

  • [Science] Forest Lizards Gained About 2°C of Heat Tolerance in a Decade

    [Science] Forest Lizards Gained About 2°C of Heat Tolerance in a Decade

    Lizards living deep in Caribbean forests can now tolerate about 2°C more heat than their counterparts a decade ago, while species that can step into cooler spots barely changed. The contrast, reported by a Yale-led team in PNAS, suggests that an animal’s habitat and its ability to regulate body temperature through behavior may influence how quickly its heat tolerance changes in a warming world.

    原論文情報 / Original Paper
    Title: “Thermal ecology drives alternate responses to rising temperatures in Caribbean anoles”
    Authors: Saúl F. Domínguez-Guerrero et al.
    Journal: Proceedings of the National Academy of Sciences (2026)
    DOI: 10.1073/pnas.2537152123

    What the team found

    Researchers led by Saúl Domínguez-Guerrero and Martha Muñoz at Yale University compared how well 16 species of Caribbean anoles, small tree-dwelling lizards, handle extreme heat. The species live on three islands: Hispaniola, Jamaica and Puerto Rico.

    The team did not set out to catch evolution in action. They were studying how behavior shapes physiology, and they expected familiar results. Back in New Haven, though, the temperature data told an unexpected story, and a colleague’s offhand remark that something looked “weird” turned into the paper.

    The measure at the center of the study is the critical thermal maximum, or CTmax: roughly, the highest body temperature an animal can tolerate before reaching a defined point of impairment in an experiment. It is a measured threshold, not a prediction of the temperature that would kill the animal. Earlier teams had measured it for the same group of species on the same islands in 2011–2013. The Yale team repeated the work in 2022–2025, giving them a before-and-after comparison across roughly a decade.

    Not every species moved the same way

    The surprise was that the shifts were not uniform. Species in open or forest-edge habitats showed little change in heat tolerance. These lizards can move into shade or onto cooler perches, which helps keep their body temperatures in a suitable range. This strategy is known as behavioral thermoregulation.

    Species deep in closed forest canopy told a different story. With far fewer ways to escape the heat, their heat tolerance rose by about 2°C over the decade. According to the researchers, thermal ecology, meaning how an animal navigates the temperatures available in its environment, predicted these changes better than evolutionary relatedness or geographic closeness did.

    Bar chart: closed-forest anoles gained about 2 degrees C of heat tolerance in a decade, open-habitat species changed little
    Figure 1. Reported change in heat tolerance over about a decade. Only the ~2 °C figure is a reported value; the other bar is schematic. Source: Yale University release on Domínguez-Guerrero et al., PNAS (2026).

    Why habitat matters

    An animal that can move into cooler surroundings may reduce its exposure to extreme heat, which lessens the pressure on its body to become more heat-tolerant. But when cooler places are scarce or hard to reach, the animal has fewer ways to avoid overheating, and stronger pressure for physiological change may follow. In the researchers’ words, behavior may help determine not only how animals respond to environmental change, but also how quickly they evolve.

    Schematic: open habitats let lizards use behavior to avoid heat, closed forest leaves heat on the body and tolerance rose
    Figure 2. Conceptual diagram of the two routes (not measured data). Source: based on the Yale University release.

    Behavior as a buffer and a brake

    This logic echoes older work. In a 2003 paper in The American Naturalist, Raymond Huey and colleagues described how thermoregulatory behavior can reduce the selection pressure for changes in thermal physiology, an effect they called the Bogert effect. A 2014 PNAS study led by Jennifer Sunday, covering 296 species of cold-blooded animals, found that heat-tolerance limits are broadly similar across latitude and elevation, that tropical species lean more on behavior such as seeking shade, and that few species have much spare heat tolerance.

    Against that background, the Yale study adds a before-and-after comparison, over a decade, of closely related lizards that differ in their opportunities for behavioral thermoregulation.

    The team also tested lizard offspring to see whether the higher heat tolerance can be inherited. The Yale release does not detail the outcome, so readers who want the answer should consult the paper itself. In follow-up work mentioned by the researchers, similar patterns appear in lizard populations in Mexico, and frog tadpoles in temperature-stable ponds also seem to be gaining heat tolerance. Those observations were described only briefly.

    Not an escape hatch

    Lead author Domínguez-Guerrero cautions that both behavioral and physiological responses have limits. “These lizards may be racing to keep up,” he said, but they may eventually hit an “evolutionary ceiling,” a point beyond which their bodies can no longer adapt. The study did not establish where such a ceiling lies; the term is the researcher’s warning rather than a measured result.

    Rapid increases in heat tolerance in some species do not mean climate warming is harmless for reptiles. They show that different species respond in different ways, and that some may be better placed to cope than others.

    Xsciens Editorial Note

    Core point: lizards that cannot escape heat gained about 2°C of heat tolerance in ten years, while species with cooler hiding places changed little.

    • Observation versus interpretation. What was measured is the upper heat limit of 16 species at two points in time, 2011–2013 and 2022–2025. A before-and-after comparison alone cannot separate genetic evolution from flexible adjustment within an animal’s lifetime, so calling the change “adaptation” is the researchers’ interpretation. The roughly 2°C figure is for closed-forest species as a group, not every species.
    • Limits. The study compares two measurement periods rather than tracking the lizards continuously, and covers 16 species on three islands, so the pattern may not hold for all animals. We worked from the university release and news coverage because the full paper was not accessible to us, so details such as sample sizes per species were not available. The Mexico lizard and tadpole findings were mentioned only briefly.
    • What to watch next. How much of the change is heritable, whether other regions and animal groups show the same split, and how far heat tolerance can rise before physiological limits are reached. The 2014 survey of 296 species found few with much spare heat tolerance.
    • From the editors — a shady patch of forest may be worth as much to wildlife as any technological fix. If the pattern holds, planners could use it to spot species with few cool places to retreat to and protect the shade, dense vegetation and cool refuges around them first, though this study did not test such measures. The warning is built in: the researchers themselves expect an upper limit, so this is no reason to relax about warming. For the rest of us, even a tree, some shade and a water dish in a backyard count as habitat for the small animals nearby.

    References for this note: Huey, Hertz & Sinervo (2003), The American Naturalist, Sunday et al. (2014), PNAS

    Sources: Yale News, Phys.org, PNAS (Domínguez-Guerrero et al., 2026)

  • A Simulated Brain Circuit Builds Its Own AI-Like Learning System, MIT Finds

    A Simulated Brain Circuit Builds Its Own AI-Like Learning System, MIT Finds

    For decades, neuroscientists have faced a nagging puzzle. Modern AI owes its power to backpropagation, an algorithm that adjusts a network’s connections by propagating error information backward through precisely coordinated connections — a computation with no clearly established equivalent in biological circuitry. Yet some studies have found that activity patterns in trained artificial networks resemble those recorded in the brain’s own visual system, which only sharpens the question of how biological neurons, using only local signals and changing synapses, might approximate something similar. A new study from MIT’s McGovern Institute for Brain Research, published October 9 in the Proceedings of the National Academy of Sciences (PNAS), offers a candidate answer: when simple neuron-like units are wired randomly and left to follow local, biologically plausible plasticity rules, they can spontaneously self-organize into circuits that learn nearly as well as backpropagation does on the tasks tested — without anyone telling them how.

    原論文情報 / Original Paper
    Title: How Biological Synapses Self-Assemble Gradient Learning
    Authors: Qianli Liao, Liu Ziyin, Yulu Gan, et al.
    Journal: Proceedings of the National Academy of Sciences (PNAS)
    DOI: 10.1073/pnas.2606245123

    A 40-Year Mismatch Between Brains and AI

    Modern deep learning owes almost all of its success to backpropagation: a network makes a prediction, compares it with the correct answer, and then propagates that error backward, layer by layer, nudging every connection weight in the direction that would have reduced the mistake. It works brilliantly, but standard implementations depend on precise mathematical relationships between forward and backward connections, plus coordinated transmission of error information through the network. There is no established evidence that cortical circuits implement anything equivalent in its conventional form.

    The puzzle deepened as deep learning matured. Comparisons between the internal activity of backpropagation-trained networks and recordings from biological visual systems have repeatedly found intriguing similarities — not identical, but notably closer than chance would predict. Such similarities do not prove the brain runs backpropagation, but they raise an obvious question: if it does not, what local process could still produce computations that end up looking so much like it?

    Local Rules, No Blueprint

    The MIT team, led by Qianli Liao, Liu Ziyin and Yulu Gan with senior authors Mark Harnett and Tomaso Poggio (and co-author Brian Cheung), approached the question differently from most prior attempts to make backpropagation “more biological.” Instead of designing a circuit by hand and checking whether it could plausibly exist in a brain, they started from almost nothing: layers of neuron-like units connected at random, with no predetermined wiring diagram at all.

    The system runs on local synaptic plasticity rules, including heterosynaptic mechanisms that let a connection change in response to activity beyond that single synapse — a more specific relative of the familiar Hebbian idea that “cells that fire together wire together.” Critically, these rules only need information available locally, at or near a given synapse (the junction where one neuron signals to the next). No unit needs to know what is happening elsewhere in the network, and no single global error signal is broadcast to the whole system.

    Diagram showing random wiring self-organizing into a learning circuit
    Source: created based on findings from MIT McGovern Institute / PNAS (2026).

    Circuits That Wire Themselves

    Left to run under those local rules, the randomly connected units did not stay random for long. Over repeated rounds of activity, the network developed small, recurring wiring patterns the researchers call “self-assembling motifs.” The model’s analysis points to a specific four-synapse motif linking the brain’s ascending, bottom-up pathways with its descending, top-down pathways — a concrete structural prediction that experimentalists can now go looking for in real neural tissue, not a pattern already confirmed to exist there.

    What these motifs do next is the more remarkable part: networks of them self-organize further, so that their combined dynamics approximate a generalized form of stochastic gradient descent — the family of optimization methods that includes backpropagation — without a hand-designed, globally coordinated algorithm directing them. In other words, before the network learned anything from its training data, it first had to build the machinery that would let it learn.

    Starting from random connectivity, local synaptic plasticity rules were enough for the network to assemble circuit motifs whose combined dynamics approximate gradient-descent-style learning.

    Holding Its Own Against Backpropagation

    A biologically plausible circuit is only interesting if it actually works, so the researchers tested their self-organized networks on image pattern recognition and classification tasks. According to MIT, the self-assembled circuits achieved performance comparable to conventional networks trained directly with backpropagation on the tasks tested. The MIT News announcement does not report specific accuracy figures or benchmark datasets — those details sit in the full PNAS paper — so the result is best read as evidence that self-organization carried no obvious cost on this particular test, not as a claim that it matches backpropagation across deep learning in general.

    That matters because it suggests biological plausibility and computational competence are not necessarily in tension. For years, biologically inspired alternatives to backpropagation have often underperformed it on real tasks. A circuit that assembles itself from random wiring and still competes with backpropagation on the tasks it was tried on narrows that gap, at least in this one demonstration.

    A Prediction Neuroscientists Can Go Test

    The researchers are careful to note that a self-organizing network learning like this does not prove the human brain learns this way — it is a candidate mechanism, not a confirmed one. But unlike many theoretical proposals in computational neuroscience, this one comes with a concrete, falsifiable prediction: if the brain really does something like this, cortical circuits should contain that specific four-synapse motif connecting ascending and descending pathways. Experimental neuroscientists can now search actual brain tissue for exactly that wiring signature, which is what turns an elegant simulation into a testable scientific hypothesis.

    Why It Matters Beyond the Lab

    If the prediction holds up, the payoff would extend in both directions. For neuroscience, a confirmed biologically plausible learning circuit would help explain how the brain adapts throughout life, and could sharpen our understanding of what goes wrong in conditions involving altered learning or memory, as well as inform the design of future brain-computer interfaces that need to interact with the brain’s own learning rules rather than fight against them.

    For AI, the lesson cuts the other way: today’s networks are built top-down, with every connection and training rule specified by engineers before a single example is seen. A system that can bootstrap its own learning machinery from random wiring and local plasticity rules hints at an alternative path — architectures that configure themselves as they encounter data, rather than ones that are handed a finished blueprint. Liao, Ziyin, Gan and their colleagues frame their result as an early step in that direction, not a finished product, but their work gives both neuroscientists and AI researchers a concrete new question to test: not just how networks learn once they are built, but how learning itself can come preinstalled.

    For more on where artificial and biological intelligence still diverge, see our earlier coverage of how MIT neuroscientists found the brain re-files information the instant you make a decision, a separate finding that is likewise reshaping how AI researchers think about adaptive, flexible learning.

    Sources: MIT News, Proceedings of the National Academy of Sciences (PNAS), arXiv preprint

  • [Science] China’s Chang’e-6 Moon Samples Reveal a Rare Magnetic Mineral — A New Lead in the Mystery of the Moon’s Lost Protective Field

    [Science] China’s Chang’e-6 Moon Samples Reveal a Rare Magnetic Mineral — A New Lead in the Mystery of the Moon’s Lost Protective Field

    • Earth’s magnetic field acts like an invisible shield, deflecting much of the charged-particle wind that streams from the Sun — but the Moon and Mars both lost a global magnetic field like this billions of years ago, and scientists still do not fully agree on why.
    • Researchers studying soil that China’s Chang’e-6 mission brought back from the Moon’s far side found a rare, naturally magnetic form of iron that had never before been confirmed in real Moon rock.
    • Using a specialized microscope, they showed that this iron can hold a stable magnetic swirl pattern — the kind of structure that can, in principle, lock in a trace of the magnetic field present when it formed.
    • This is an early-stage finding: it does not by itself settle the long-running debate about the Moon’s lost magnetic field, but it gives scientists a new kind of physical evidence to test their ideas against.

    A missing magnetic shield, and a Moon that might hold a clue

    Right now, a layer of invisible magnetism generated deep inside Earth is quietly doing one of the most important jobs for life on this planet: deflecting much of the charged-particle wind that streams from the Sun, known as the solar wind, so it does not gradually strip away our atmosphere. Mars does not have this shield anymore. Neither does the Moon. Both are thought to have had one billions of years ago, generated by a “dynamo” — a churning flow of molten or partly molten metal inside the planet that behaves like a natural electric generator. Figuring out when and why each dynamo switched off is one of the open questions in planetary science, partly because it is also a preview of how a rocky world’s interior can change over very long timescales.

    The Moon is a convenient place to study this, because unlike Mars, it is close enough that spacecraft have brought real samples home. In 2024, China’s Chang’e-6 mission did something no previous mission had: it collected soil from the Moon’s far side, from the Apollo basin inside the giant South Pole–Aitken impact basin. A team at the Hefei Institutes of Physical Science (HFIPS), part of the Chinese Academy of Sciences, has now reported a new discovery hiding inside that soil, published online September 16, 2026, in the Proceedings of the National Academy of Sciences (PNAS).

    An iron that “shouldn’t” exist — until an asteroid hit made it

    The team, led by Prof. Haifeng Du, was looking at tiny beads of impact glass: droplets of rock that melted and then froze solid within moments after an asteroid or other object slammed into the Moon’s surface. Scattered through that glass, they found nanometer-scale particles — far smaller than a speck of dust, ranging from under 10 nanometers to about 140 nanometers across — made of a form of iron called gamma-iron, or γ-Fe for short.

    Gamma-iron is not exotic chemistry; it is simply metallic iron arranged in a different crystal structure than the form most people picture. At ordinary pressure, pure iron normally settles into a structure called alpha-iron as it cools, and that is the arrangement found in most familiar iron-rich rocks. Gamma-iron is normally only stable over a high-temperature range, roughly 912°C to 1,394°C, and should flip back into the ordinary form well before reaching room temperature. Finding it “frozen” in place at room temperature, inside real Moon soil, was unexpected. The researchers propose that the extremely rapid cooling that follows a high-speed impact — far faster than rock cools under ordinary conditions — combined with the small amount of trace elements in the particles, trapped the iron in this unusual structure before it had a chance to transform.

    Among 50 iron particles the team examined across two impact-glass samples, γ-Fe was the dominant crystal phase: it accounted for roughly 64% of the particles analyzed in one sample and 76% in the other. The particles also contained very little nickel, below 0.8% by weight, which helps distinguish them from the iron-nickel alloys already known from meteorites and other Moon rocks.

    Bar chart showing the share of analyzed iron particles identified as gamma-iron in two Chang’e-6 impact-glass samples: 64% and 76%.
    γ-Fe was the dominant iron phase in both impact-glass samples the team analyzed in detail.

    Reading a magnetic memory under the microscope

    Finding an unusual mineral is interesting on its own, but the more important part of the discovery is what the iron does magnetically. Using a specialized imaging technique called off-axis electron holography — a way of directly visualizing magnetic structure at a scale of billionths of a meter, inside an electron microscope — the team found that γ-Fe particles larger than about 100 nanometers settle into a stable “single-vortex” magnetic state: the magnetization inside the particle curls around in a stable loop, rather than pointing in one simple direction.

    That stability matters. A magnetic structure that holds its shape reliably, and responds in a predictable way to outside magnetic fields, is the kind of structure that can carry what scientists call remanent magnetization — a lasting trace of the magnetic field present when the material formed and cooled. In principle, that makes it somewhat like certain magnetic minerals in Earth’s own rocks, which preserve a record of our planet’s past magnetic field. Dr. Long Li, one of the researchers on the team, put it this way:

    “This tiny magnetic fossil may help us better understand the Moon’s ancient magnetic history.” — Dr. Long Li, Hefei Institutes of Physical Science

    It is worth being precise about what has, and has not, been shown so far. The team has demonstrated that γ-Fe exists in real lunar material, that it forms a stable magnetic vortex state, and that this state is capable of recording magnetization. What is not yet established is which magnetic field any particular particle actually recorded — the Moon’s ancient, planet-wide dynamo field, or a brief, local field generated by the heat and shock of the very impact that formed the glass bead around it. Telling those two signals apart is exactly the next step the researchers say this discovery opens up.

    A genuinely unsettled debate about the Moon’s lost magnetism

    Scientists already have evidence the Moon once had a working dynamo, because older Moon rocks brought back decades ago by the Apollo missions carry their own trace of ancient magnetism. Exactly how strong that field was, and especially how long it lasted, is still disputed — and this is a real, ongoing scientific disagreement rather than a settled fact with one outlier view. Paleomagnetic measurements of Chang’e-6 basalt reported a field of roughly 5 to 21 microtesla as recently as about 2.8 billion years ago, a result some researchers read as pointing toward a longer-lived dynamo. Other researchers, using different measurement methods on Apollo-era samples, have argued that the evidence for a long-lived global field is weaker than it first appeared, and that the core dynamo may have been largely confined to the Moon’s first 140 million years or so, with anything recorded afterward reflecting brief, impact-generated fields rather than a steady, planet-wide one.

    Timeline diagram comparing two rival hypotheses for when the Moon’s internal dynamo stopped generating a global magnetic field.
    Two competing timelines for when the Moon’s ancient dynamo shut off — a debate γ-Fe could eventually help settle.

    This is where γ-Fe could eventually contribute. Because it forms under different conditions than the more common alpha-iron, and appears specifically in impact-formed glass, comparing what it and other magnetic minerals “remember,” across many more samples, may eventually help researchers test whether a given signal reflects the Moon’s global dynamo, an impact-generated field, or some mixture of the two.

    Still early days: what comes next

    The team has not announced a new, confirmed timeline for when the Moon’s dynamo turned off, and γ-Fe on its own cannot provide a precise date for that event. What they have shown is that this magnetic mineral exists in real lunar material, that it can hold a stable magnetic state at the nanometer scale, and that it is a promising new tool for a field that has had to work with a frustratingly small and ambiguous set of physical clues. More broadly, understanding how a rocky world like the Moon or Mars generates, and eventually loses, its own internal magnetic field is part of a wider effort to understand what shapes a planet’s long-term climate and atmosphere — knowledge that also feeds into how scientists assess the conditions on rocky planets orbiting other stars. A small, newly identified mineral will not settle that larger question by itself, but it gives researchers one more concrete thread to pull on.

    Sources: Du, H. et al., “Magnetic vortex state of natural lunar γ-Fe,” PNAS (2026) · Hefei Institutes of Physical Science, CAS · ScienceDaily · The Brighter Side of News