Why Most Species Fail to Take Root — Even When the Door Is Open
Even when 100 species arrive at once, only about 3 survive. Not because of predators or climate — but because ecosystems resist newcomers with silent, invisible force.
This is the counterintuitive insight from a new model of ecological fitting, a process where species interact not because they evolved together, but because their traits happen to align in a new environment (Arthur et al., 2026). Think of it as biological serendipity: a plant from South America thrives in Hawaii not because it adapted, but because its root system accidentally pairs well with local soil bacteria. These chance matches explain how invasive species establish, how novel ecosystems form, and even how life recovers after mass extinctions.
But the model reveals something deeper: ecosystems aren’t just full — they’re guarded. The more diverse and populous an ecosystem, the harder it is for any new species to invade, even if it’s perfectly suited. And when entire communities of species colonize a barren landscape — like on Ascension Island in the Atlantic — only a tiny fraction persist. The rest vanish, not from competition, but from a hidden mathematical logic of interdependence.
This isn’t just theory. It reshapes how we think about conservation, restoration, and the future of life on a changing planet.
The Science
The study builds on the concept of ecological fitting, first proposed by ecologist Daniel Janzen in 1985 after observing complex webs of interactions among species in Costa Rica that had no shared evolutionary history (Janzen, 1985). He realized that just because two species interact — say, a bird eats a fruit — doesn’t mean they co-evolved. Sometimes, it’s just luck. One species’ trait happens to work with another’s, like a key fitting a lock it wasn’t made for.
This idea has since become central to understanding invasive species, host-parasite jumps, and the rise of so-called novel ecosystems — environments like urban parks or reclaimed mines where species from around the world coexist without historical ties (Hobbs et al., 2013). But until now, there was no general model to predict when and why ecological fitting succeeds.
Arthur, Nicholson, and Wilkinson (2026) fill that gap with a mathematical framework based on a modified version of the classic Lotka-Volterra equations, which describe how species populations change over time due to growth, competition, and interactions.
Their key innovation is twofold. First, they remove the idea that any species can survive alone. In their model, every species depends on others for survival — no intrinsic growth rate, no solo existence. Second, all species share a single carrying capacity: space, light, nutrients, water. If one species grows, it directly reduces the resources available to all others, even if they don’t directly compete.
The model is defined by:
where:
- $N_i$ is the population of species $i$,
- $N = \sum_i N_i$ is the total population,
- $J_{ij}$ represents the interaction strength from species $j$ to $i$ (positive for mutualism, negative for competition),
- $\mu$ sets the overall carrying capacity ($\mu = 0.01$ in simulations),
- and $J_{ii} = 0$, meaning no self-interaction.
The interaction matrix $J_{ij}$ is random: with probability $\Theta$, the interaction is zero (no link), otherwise it’s drawn from a normal distribution with mean 0 and variance 1. This captures the idea that most species don’t interact, but when they do, the effects are unpredictable.
The model is studied in two scenarios:
- Ecological colonisation: many species arrive simultaneously on a barren landscape (inspired by Ascension Island).
- Ecological invasion: a single new species attempts to join an established ecosystem.
Both are solved numerically, with populations below $10^{-6}$ set to zero — a practical threshold for extinction.
What They Found
Colonisation: From 100 Species to Just 3
When 100 species are dropped into an empty world, you might expect a rich, diverse ecosystem to emerge. But the simulations show otherwise.
Figure 2 shows the results across thousands of simulations. Even when the starting pool is large — up to 100 species — the average final diversity is shockingly low. For most parameter values, it’s less than 4. The maximum observed diversity? Just 10, and that was rare.
Why?
The model reveals a process the authors call virtual group selection. Species don’t survive as individuals, but as networks — pairs, trios, or small clusters that mutually support each other. These networks compete not just for resources, but for dominance in the shared environment.
Consider two mutualistic pairs: one with weak positive interactions, another with strong ones. Even if they don’t directly interact, they compete for the same carrying capacity. The stronger pair grows faster, consumes more of the shared resource, and indirectly suppresses the weaker one — not through conflict, but through crowding.
This creates a size-diversity trade-off. Larger groups have more interactions, but dilute their per-capita benefit. Smaller, tightly coupled groups can outcompete larger, looser ones. The math shows that the winning group maximizes $p_b r_b$, where $p_b$ is its share of the total population and $r_b$ is its internal interaction strength. Small groups with high $r_b$ win.
Final diversity rarely exceeds 4, even with 100 starting species
| Label | Value |
|---|---|
| 10 | 2.8 |
| 20 | 3.1 |
| 50 | 3.4 |
| 100 | 3.6 |
Chart 1: Final diversity rarely exceeds 4, even with 100 starting species
| Starting Species (M) | Avg. Final Diversity (D) |
|---|---|
| 10 | 2.8 |
| 20 | 3.1 |
| 50 | 3.4 |
| 100 | 3.6 |
Data from Figure 2(b), averaged over 1000 simulations per M, Θ = 0.25.
The result is a kind of ecological winnowing. Out of hundreds of possible combinations, only one small network prevails. The rest go extinct — not because they’re unfit, but because they’re out of sync with the emergent structure.
Invasion: The Invisible Barrier
Now consider a single invader trying to join an established ecosystem.
The model identifies a barrier to invasion:
The left side is the invader’s effective growth rate, determined by its interactions with existing species. The right side is the barrier, set by the total population $N$ and the carrying capacity parameter $\mu$.
In simple terms: the bigger and more populous the ecosystem, the harder it is to invade.
Why? Because $\mu N$ grows with $N$, while the invader’s potential benefit — a sum of random interactions — has diminishing returns as diversity increases. For large $D$, the variance of the invader’s growth rate scales as $(1-\Theta)/D$, so the chance of a strong positive match drops sharply.
The probability of successful invasion is:
This is super-exponentially small. Double the population, and the invasion chance doesn’t just halve — it plummets by orders of magnitude.
Invasion success drops sharply with ecosystem size and diversity
| Label | Value |
|---|---|
| 50 | 0.01 |
| 100 | 0.00001 |
| 150 | 1e-10 |
Chart 2: Invasion success drops sharply with ecosystem size and diversity
| Total Population (N) | Diversity (D) | Approx. Invasion Probability |
|---|---|---|
| 50 | 3 | ~10⁻² |
| 100 | 5 | ~10⁻⁵ |
| 150 | 7 | ~10⁻¹⁰ |
Calculated from Eq. 6, Θ = 0.5, μ = 0.01.
Figure 4 confirms this: in simulations, invasion failure dominates, especially in high-N, high-D systems. When invasions do succeed, they often cause rearrangement — one or more native species go extinct — or, surprisingly, total collapse of the entire ecosystem.
Only in low-diversity systems does the invader simply add to the network without disruption.
Why This Changes Things
The Diversity-Stability Debate, Resolved?
For decades, ecologists have debated whether more diverse ecosystems are more stable. Some argue yes: more species mean more redundancy, so the system can absorb shocks. Others say no: more interactions create more pathways for collapse.
This model suggests a clear answer: yes, diversity stabilizes — against invasion.
High-diversity systems aren’t just richer — they’re harder to disrupt. Their sheer size and interdependence create a barrier that filters out most newcomers. This isn’t just resistance to invasion; it’s a form of ecological inertia.
But there’s a twist: when that barrier is breached, the consequences can be catastrophic. The model shows that successful invasions often trigger total extinction — not gradual change, but sudden collapse. This mirrors real-world cases like the introduction of brown tree snakes to Guam, which wiped out most native birds.
So diversity brings stability, but also fragility — a paradox familiar from financial networks or power grids.
Rewriting the Story of Recovery
The implications go beyond today’s ecosystems. The model offers a new lens on Earth’s deep history.
After mass extinctions — like the one that killed the dinosaurs — the planet doesn’t just refill empty niches. It reboots. Survivors, few in number, compete to form new networks. This is ecological colonisation on a global scale.
The model suggests that each recovery event selects not just species, but networks. And because larger starting pools lead to slightly more diverse and productive outcomes, repeated cycles of extinction and recovery could drive a slow, long-term increase in biodiversity — a process the authors link to sequential selection with memory (Arthur and Nicholson, 2022).
In this view, mass extinctions aren’t just setbacks — they’re evolutionary resets that, over millions of years, may have helped build the rich biosphere we see today.
The Illusion of the Empty Niche
We often assume that if a niche is empty, a species will evolve or invade to fill it. But this model challenges that.
Even when niches seem open, the structure of existing ecosystems — their population size, diversity, and interaction patterns — can block entry. A species might be perfectly adapted, but if it doesn’t have strong enough positive interactions with the right partners, it fails.
This explains why many introduced species fail, even in seemingly ideal conditions. It also suggests that promiscuous species — those that can interact with many others — are more likely to succeed, as seen in real-world invasions of legumes that form flexible partnerships with soil bacteria (Le Roux et al., 2017).
What’s Next
Caveats and Open Questions
The model is abstract. It assumes random interactions, equal species weights, and a single shared resource. Real ecosystems are more complex: species vary in size, mobility, and impact; resources are multidimensional; evolution can reshape interactions over time.
The authors acknowledge this. Their goal isn’t to simulate reality, but to reveal principles — the hidden rules that govern ecological fitting.
One key question: how do results change if species can evolve? In earlier work, the same team studied a version with mutation (Arthur and Nicholson, 2017). Adding evolution could allow invaders to adapt, or residents to resist — potentially softening or sharpening the invasion barrier.
Another: what if the carrying capacity isn’t shared? Trees compete for space, but birds depend on trees. More realistic models might layer multiple resource constraints.
Applications: Restoration, Biosecurity, Terraforming
Despite its simplicity, the model offers practical insights.
For ecological restoration, it suggests that simply adding species to degraded land may fail — not because the environment is hostile, but because the network isn’t ready. Success may require introducing groups of mutually supportive species, not individuals.
For invasive species management, it reinforces the importance of early detection. Once an ecosystem is diverse and populous, it’s naturally resistant — but if an invader slips through, it could trigger collapse. Prevention is far cheaper than cure.
And for terraforming — the sci-fi dream of making other planets habitable — the model is sobering. Even if we seed Mars with 100 species, only a handful may survive. The rest will vanish, not from cold or radiation, but from the silent logic of ecological fitting.
In a world of accelerating change, where species are moving, ecosystems are fragmenting, and climates are shifting, this model reminds us that nature isn’t just a collection of organisms. It’s a web of dependencies, shaped by chance, history, and invisible mathematical forces. Some doors are open. Others are locked — not by locks, but by numbers.