The first Martians won't be human for long.

The first Martians won't be human for long.

Not in the sense of cyborgs or genetic engineering, though both are probably coming. In the purely Darwinian sense: a founding population of a few thousand people, isolated on a planet with higher radiation, lower gravity, and a completely different seasonal rhythm, will begin to diverge from the Earth population the moment the last shuttle docks. It won't be dramatic. It will be invisible at first: a few alleles drifting to fixation here, a slightly different selective pressure on bone density there. But population genetics doesn't care about drama. It cares about effective population size and time, and on those axes, Mars is a speciation machine.

The mathematics of this are laid out in the Hardy-Weinberg equilibrium framework, the null hypothesis that a population in the absence of evolutionary pressure holds its allele frequencies constant across generations. Hardy-Weinberg is useful precisely because real populations never satisfy it. Every violation of its assumptions (mutation, genetic drift, migration, selection) is a signal that evolution is happening. On Mars, all four forces operate simultaneously, and three of them push in the direction of divergence. The question isn't whether Earth and Mars populations will become genetically distinct. The question is how long it takes, and what the threshold for "distinct species" even means when we can sequence entire genomes.

The threshold question is harder than it sounds. Biological species concepts rely on reproductive isolation, but humans don't have a reliable isolating mechanism short of being on different planets, which is exactly the scenario we're modeling. Genetic distance metrics like F_st (fixation index) give us a continuous measure of divergence, but the speciation "event" is not a discrete threshold crossing. It is an accumulation of incompatibilities. The four forces of evolution (mutation, drift, migration, and selection) each contribute differently to this accumulation, and they interact in ways that forward-time simulation is uniquely positioned to capture. That is the core methodological bet of this project: simulate the process forward, track the metrics, and ask when the divergence crosses thresholds that have historically corresponded to species-level distinction on Earth.

The project started as a single scenario: pure neutral drift, no selection, no migration. A founding population of 1,000 colonists leaves Earth. Generations pass. How different do they get just from random allele frequency fluctuations in a small, closed population? The answer from the SLiM simulations was unambiguous. Neutral drift alone produces species-level F_st divergence in roughly 7,000 to 10,000 years under realistic effective population sizes. That number has a wide confidence interval, and it is sensitive to assumptions about generation time, but the order of magnitude held across parameter sweeps. Isolation plus time, even without any selective pressure, is enough.

The more interesting question, and the one that turned a proof-of-concept into a full simulation framework, was what happens when you add selection. Mars presents a suite of specific challenges that Earth never did: chronic low-dose ionizing radiation from galactic cosmic rays, reduced gravity affecting bone and muscle metabolism, UV exposure through a thinner atmosphere, and a circadian disruption that probably runs deeper than jet lag. Directional selection for Martian adaptation shortens the divergence timeline considerably. In the second simulation scenario, alleles that confer radiation resistance or altered bone remodeling reach fixation in the Martian population within tens of generations, fast enough to matter on human timescales. The third scenario is the most complex: selection with periodic gene flow, modeling the effect of resupply missions and population exchanges. Even modest gene flow substantially slows divergence, which has obvious policy implications for anyone thinking about long-term interplanetary civilization.

The framework grew to track three divergence metrics simultaneously: F_st, migration coefficient, and a reproductive isolation index derived from genetic incompatibility accumulation. Each metric tells a different part of the story, and they don't always agree on when speciation "happens." That disagreement is itself a finding. It reflects the genuine ambiguity in what speciation means for a species that can still make deliberate choices about who it allows to mate with whom, a theme the third article in this series takes head-on.

The biggest unresolved question is the supply-side of selection pressure: we don't actually know which specific traits will be under positive selection on Mars. The simulation parametrizes radiation resistance as a single quantitative trait, but the real biology is polygenic and poorly characterized even in Earth populations. The same uncertainty applies to bone density, immune function in a low-diversity microbial environment, and cognitive adaptation to social isolation at planetary scale. The divergence timelines the framework produces are therefore lower bounds on what directional selection can do. If the actual Martian adaptive landscape is richer than the model assumes, speciation happens faster.

The second open question is normative rather than scientific: at what point, if any, does the anticipated divergence create an obligation to intervene, or to deliberately not intervene? Gene flow is a speciation brake. A civilization that wants to keep Earth and Mars populations genetically unified could, in principle, enforce it through migration policy, or eventually through germline editing. A civilization that is indifferent to divergence, or that actively wants two distinct branches of the human lineage, would make different choices. The simulation can tell you what each policy implies for divergence timelines. It cannot tell you which policy is right. That is the question the modeling hands off to ethicists, and the reason this project sits at the intersection of computational biology and something much older than either.