How do you save a bird with fewer than 60 individuals? We built a model to find out
Imagine you are responsible for the survival of a species. There are fewer than 60 of them left in the wild. They live across three countries, in river valleys being rapidly transformed by dams, agriculture, and human settlement. You have limited funding, a small team, and the knowledge that every decision you make — or fail to make — will reverberate across what remains of a wild population.
Where do you start?
This is not a hypothetical. It is the reality facing conservationists working to protect the White-bellied Heron, one of the most critically endangered birds on Earth. And it is the question that drove our recent study, published in Avian Conservation and Ecology.
The problem with guesswork
Conservation decisions are often made with incomplete information. Which intervention matters most? Is it better to protect nesting sites, improve survival of adults, or invest in a captive breeding programme as a safety net? Without a clear way to test these alternatives, the temptation is to do everything — which is rarely possible with limited resources — or to rely on intuition.
We wanted something more rigorous. We wanted to be able to say: given what we know about this population, this is what the numbers look like under different scenarios.
Building a model for an extraordinary bird
To do that, we built what ecologists call an integrated population model (IPM) — a statistical framework that pulls together multiple streams of data simultaneously. In our case, that meant combining annual population counts, breeding success data from nest monitoring, and estimates of adult and juvenile survival, all collected over more than a decade of fieldwork in Bhutan.
By combining these data streams rather than analysing them separately, the model gives more precise and internally consistent estimates of what is actually driving population change.
We then extended this into a Bayesian population viability analysis (BPVA) — essentially a way of running thousands of simulated futures for the population and seeing how likely different outcomes are under different conditions. Bayesian methods are particularly well-suited to this kind of work because they properly account for uncertainty: we do not pretend to know things more precisely than we actually do.
What the model found
The results are confronting. Without any change in current conditions, our simulations suggest the wild White-bellied Heron population in Bhutan faces a high probability of continued decline. The model is not predicting imminent extinction — the birds are still there, still breeding — but the trajectory is in the wrong direction, and the margin for error is narrow.
The analysis also let us ask: what if things changed? We modelled a range of conservation scenarios — increasing nest survival through active protection, reducing adult mortality, managing disturbance at key sites — and examined how each one affected the population’s prospects over the coming decades.
The findings pointed clearly toward the value of improving juvenile and adult survival. Interventions that keep more birds alive between breeding seasons — reducing mortality from disturbance, collision, and other human pressures — have a proportionally larger effect than improvements to breeding success alone. This matters because it tells us where to focus.
The model also reinforced the importance of the ex-situ (captive) population maintained by the Royal Society for Protection of Nature in Bhutan. In a population this small, chance events — a flood, a disease, a single bad breeding season — can have outsized consequences. The captive population acts as an insurance policy: a reservoir of individuals that can, if needed, contribute to the wild population’s recovery.
Why models matter in conservation
There is sometimes scepticism about mathematical models in conservation. Real species are complicated. Real habitats are messy. Can equations really capture what matters?
In our experience, the value of models is not that they give you a perfect prediction of the future. They do not. The value is that they force you to be explicit about what you know, what you do not know, and what assumptions you are making. They help you see which uncertainties matter most — and therefore where more data would change decisions. And they give you a common language for discussing alternatives with managers, policymakers, and funders.
For a species as rare as the White-bellied Heron, a model that synthesises a decade of painstaking field data into actionable guidance is worth every line of code.
What comes next
This study establishes a baseline. We now have a quantitative picture of where the Bhutan population stands and what is most likely to help it. The next steps involve refining those estimates as more field data accumulates, extending the model to include populations in India and Myanmar, and working with partners on the ground to translate the findings into specific management actions.
The White-bellied Heron does not have the luxury of waiting for perfect information. But armed with the best available evidence — and a model that honestly represents what we know — we are in a better position to act.
The full study — “Investigation of effective conservation measures for the critically endangered White-bellied Heron Ardea insignis using an integrated population model and Bayesian population viability analysis” — is published in Avian Conservation and Ecology, 20(1):19 (2025) and is freely available at doi.org/10.5751/ACE-02833-200119.
Authors: Indra Acharja, Yuji Okahisa, M. Clay Green, Thinley Lhendup, Sonam Tshering, Thinley Phuntsho, Tshering Tobgay, Lungten Lungten, Jigme Tshering, Sangay Leki.
Comments