The Animal Welfare Department (AWD) at Rethink Priorities supports high-impact strategies to help animals, especially where suffering is vast and largely neglected. Therefore, one of our focus areas is wild animal welfare (WAW), where uncertainty about tractability makes identifying cost-effective interventions particularly challenging. While much of the current WAW work rightly focuses on academic field-building (see Elmore & McAuliffe, 2024), it is worth determining whether there are viable, near-term interventions that are already available or close to implementation. 

With this goal in mind, we have developed the Wild Animal Welfare Intervention Database (WAWID). This project evaluates an array of interventions that may be promising for improving WAW in the (relatively) near term, evaluating them relative to criteria of interest to funders, advocates, researchers, and potential implementers across the WAW space.

The WAWID is available here: 

The landing page includes a full list of the interventions and evaluation criteria. This report explains how we developed the WAWID, what we think you can learn from it, and suggest some future directions for this work (conditional on funding). A future report will provide some descriptive statistics.

How we developed the WAWID

We launched this project in September 2025. We began with a short list of interventions that have frequently been discussed in the effective animal advocacy movement (e.g., fertility control for rodents, bird-safe glass, more humane pesticides) and a short list of criteria by which to evaluate them (e.g., number of animals directly benefited, backfire risk for non-target animals, promise for coalition building). We then circulated this list among people who work in this space, soliciting advice about other interventions and evaluation criteria worth including. 

We were able to include all the evaluation criteria that emerged during that initial process. (This is not to suggest that our list of criteria is exhaustive: there are bound to be others that would be of interest to some stakeholders.) We do not adopt any one stance on how to weigh the evaluation criteria against one another. While we do later explain how different worldviews would prioritize different criteria, we do not endorse any specific worldview nor suggest how best to cope with uncertainty over them. 

The people we consulted suggested many possible interventions that might be worth investigating—far more than we could pursue given the limited time and resources available. We decided which interventions to include based on several factors, such as:

• Level of interest in the intervention in WAW circles
• Whether we thought it was plausible that the technology required for the intervention would be ready to go in the next decade
• Whether the intervention is such that a relatively small number of people could implement it, given likely funding scenarios and their implications for available staffing
• Whether there are particularly serious optics risks for the movement if it pursues the intervention
• Overlap with other included interventions
• Availability of a researcher with an interest in that intervention

Our list of interventions is intentionally diverse; it isn’t meant to be guided by a single set of philosophical or empirical assumptions (which is one reason why none of the above factors was decisive). Examples include:

• Pigeon fertility control. This shallow examines lobbying for the use of nicarbazin-based contraceptive baits to manage urban feral pigeon populations, focusing on commercially available products (such as OvoControl-P in the US) as a more humane alternative to culling and lethal pest control methods.
• Mating disruption in rice stem borers. Mating disruption uses species-specific sex pheromones to prevent Yellow Stem Borers from finding mates in rice fields, offering a non-lethal alternative to insecticides that has been associated with 40–56% fewer pesticide applications in Indonesia.
• Wild animal welfare considerations in urban planning. This shallow evaluates an intervention to encourage city councils to adopt high-level commitments to consider wild animal welfare in urban planning, finding that while such ordinances are likely feasible, it is unclear whether they would improve wild animal welfare significantly and could set it back.

No one should interpret the list as a statement about RP’s values or priorities, nor ours as researchers. We acknowledge that there is arbitrariness in the list’s contents, some of which was inevitable, as there are indefinitely many interventions that could sensibly be considered and no obviously optimal set to consider.

We gave researchers 20 hours to complete each shallow. We told them to focus on the following:

• Scale: Research a potential problem for wild animals, quantifying its scale and severity insofar as possible.
• Levers: Identify possible changes that, if made, would address this problem;
• Robustness: Assess whether those animals (and, ideally, other affected animals) would be better off on net due to those changes; and
• Non-Movement Resources: Determine the extent of the non-movement resources available to bring about these changes.

We gave these instructions with one set of stakeholders in mind. For some, the primary utility of the database is as a way to screen for potentially cost-effective ways of reducing the suffering of wild animals in the near-term. Scale, Levers, Robustness, and Non-Movement Resources shed some light on the prospects for cost-effectiveness. If the scale is modest, then an intervention is unlikely to be cost-effective unless it’s possible to redirect a lot of non-movement resources toward it. If an intervention isn’t robust, then our uncertainty about its net effects is– at least for the ambiguity-averse–  a discount on its cost-effectiveness (among other things). If there aren’t any non-movement resources to rally, then the scale needs to be large enough, and the robustness sure enough, to offset the required spend of movement resources. 

For other users, however, the database largely serves as a tool to identify interventions that help attract talent to the movement, increase sympathy for wild animals among powerful stakeholders, or serve as a persuasive proof of concept that we can responsibly intervene in nature. So, we had researchers evaluate all interventions relative to criteria like: 

• Movement-Building Value: How much does this intervention strengthen the broader wild animal welfare movement and field? Make sure to consider reputational impacts.
• Non-Movement Fundraising Prospects: How likely is this intervention to attract non-movement funding? And how much funding?

We often gave researchers fairly broad topics and told them to scope the problem down based on data availability. For instance, it’s impractical to research something as general as “a ban on some insecticides.” But it may be feasible to look into the prospects for banning a certain class of neonicotinoids in the United States, for instance, and then investigate how many insects neonicotinoids affect in the US, the severity of those impacts, etc.

All the shallows and scoring sheets were reviewed by three people: the editor and two researchers. Between the time devoted by those individuals and the original researchers' responses, this probably brought the total amount of time spent on each shallow/scoring sheet pair up to around 30 hours. The exception to this rule is the scoring sheets where we did not commission an original shallow. In some cases, there was enough existing work that the scoring sheet alone seemed sufficient. Between the original scoring and the review process, people probably spent five to ten hours total on each scoring sheet of this type.

Initial observations

In the sequel to this report, we will assess how the relative promise of interventions depends on strategic beliefs and moral values, as well as how many interventions seem promising in absolute terms. For now, we limit ourselves to some anecdotal observations we made while reviewing the shallow assessments:

There's something for everyone, but nothing robust to all worldviews. Something like fertility control for pigeons, for example, is going to be very attractive to those who focus on creating coalitions: it’s easy to see how it might garner support from the public, which could lead to various partnerships with policymakers, urban wildlife managers, and others. However, as the shallow notes, “a meaningful fraction of wildlife contraception interventions for wild animals could lead to a counterfactual decrease in welfare.” Insofar as we want to avoid this risk, fertility control measures may be far less appealing. Likewise, the number of individuals managed during spongy moth outbreaks is enormous; so, the expected value of intervention to reduce their populations, thus preventing potentially painful suffering through management practices, is enormous too. However, some people will be wary of investing energies in animals of uncertain sentience; moreover, there are potential reputational risks of directing resources toward them. So, we should again expect a divide in the intervention’s appeal. 

High value of information for further research. As one contributor put it in a note on a draft version of their shallow: “The issue [under discussion] is very complicated. Despite having already spent ~25 hours on this shallow, I found a number of contingencies that drastically changed my mind in the last few hours of writing. I expect that many more such contingencies exist and, as such, my conclusions and recommendations in this report are very low confidence.” 

Further research will be costly and will take time. There may be cases where more desk research could help, as it takes time to develop a sufficiently rich understanding of the many factors at play in assessing an intervention, ranging from foundational ecological questions to the cultural, economic, and political forces that preserve the status quo. But we suspect that empirical research will typically be essential. For example, removing or retrofitting blocked culverts would let migratory fish move upstream and downstream more easily. However, as the author of that shallow notes: 

• Most of the individuals in the Centrarchidae, Cyprinidae, and Fundulidae families in the United States are omnivorous. This means that many of them will eat other fish and macroinvertebrates, as well as the eggs of other fish species. Successfully reconnecting their habitats will mean that complex predation dynamics will play out, which may involve pain, change average lifespans, alter the number of individuals coming into existence, and tip the balance of accessible nutrients in resource-scarce environments.
• Populations on either side of a barrier can reach separate equilibria. Reconnecting them does not guarantee a return to a cohesive whole; it may create a collision of these established populations.
• Even where predation dynamics represent less of a concern from a welfare perspective, it is possible that expanding the individuals’ habitat will not be a net win. For welfare expansion to be good from a net welfare perspective, the previously inaccessible upstream habitat should be of sufficiently high quality to sustain future generations who now have access to it; the previously separated downstream habitat should be of sufficiently high quality to not pollute the upstream habitat. 

They conclude by saying:

A key question is whether, assuming welfare is net positive in unfragmented ecosystems, we can assume that reconnecting them is a net win "on average." My best answer is that I cannot tell, and I think answering this question would be difficult. This is not to say that culvert retrofits cannot represent a net win for the individuals whose populations are affected—even when the intervention is a ripply one—but my impression is that it 1) would be hard to tell if this will occur on the basis of desk research alone, 2) would need to be evaluated on nearly a case-by-case basis, and 3) is not the priority of the environmental groups who are proponents of culvert retrofits.

In other words, the relevant research would take time, it would be difficult to generalize from it, and we probably can’t expect the research to occur without significant movement investment. Many other interventions in the database face comparably daunting empirical uncertainties. Those enthusiastic about wild animal interventions may feel stuck: on the one hand, they feel urgency to act now; on the other, they recognize that it may take decades of concentrated effort to feel confident about a given intervention.

Limitations

Beyond capacity limitations to assess even more interventions, the WAWID has many limitations. Here are some of them.

There are value judgments implicit in the selection of the evaluation criteria. For instance, the risk assessments are unbalanced, as we didn’t assess the risk of reducing the length or number of positive lives. The practical import of this omission depends on how much uncertainty readers have about the overall quality of wild animals' lives. In our experience, people who prioritize wild animal welfare tend to make pessimistic empirical assumptions about the balance of suffering against happiness in nature. Some, however, are more optimistic or counsel against sweeping generalizations given the dearth of direct empirical evidence. The latter camp is more likely to benefit from future research that could assess the likelihood interventions in this database may curtail lives worth living (e.g., Browning & Veit, 2023 Cuddington, 2019). 

No attempt to address interrater reliability. While the scoring assessments were reviewed by multiple people to ensure standardization and address inconsistencies between the shallows and the scoring sheets, only the researcher completed an independent assessment of each criterion. We are unable to quantify the interrater reliability of the scores. We expect that there is room for considerable disagreement on many assessments, given the time constraints and the generally poor evidence base available on which to make judgments about many matters.

No attempt to mitigate systematic bias. Even when raters agree, this may be due to a systematic bias that we made no attempt to mitigate. Given the time limitations, we did not require researchers to justify their more subjective judgments beyond a few short notes. In our estimation, for example, researchers tended to be rather optimistic about the movement-building value of many interventions. This may be due to selection bias in who becomes interested in wild animal welfare; it may be due to the interventions we chose.

Future Directions

In our next report on the WAWID, we will provide some descriptive statistics. RP is also open to extending this database insofar as there is clear user demand. Here, for instance, are a few enhancements that could be valuable: 

Cost-effectiveness template: Incorporate an approach that assumes an organization budget of a certain size and allow people to plug in reasonable estimates for the probability of impact, the scale of impact, and so on. Complete for each shallow.

Expert input: Spot-check several of the interventions to see how much our overall assessment changes after, say, three interviews with domain experts. This exercise would also give us some sense of how much we should value generalist research versus specialist research in these areas.

Expanded list of interventions: We didn't have time to assess several interventions that are taken seriously within the wild animal movement. Examples include conservation of large herbivores and banning glue traps. 

Expanded list of evaluative criteria: Some stakeholders may be interested in evaluative criteria that this project doesn’t include. A future version could assess all interventions by those standards. 

If there is an extension that would be valuable to your work as a funder or advocate, please email [email protected] or [email protected]. 

Acknowledgements

This report is a project of the Animal Welfare Department at Rethink Priorities. It was written by Bob Fischer and William McAuliffe. The researchers on this project included Kristian Brevik, Nicolas Delon, Simon Eckerström Liedholm, Joe Gottlieb, Mal Graham, Shloka Janapaty, Tristan Katz, Katrina Loewy, Benjamin Ransom, Jake Schmiess, Ben Stevenson, and the Screwworm Free Future team. Thanks to Thais Jacomassi for copyediting; thanks to Ula Zarosa for coordinating. If you like our work, please consider subscribing to our newsletter. You can explore our completed public work here.