Selecting strategies for high-risk island rodent eradications

1 Introduction

Invasive rodent eradication has become a reliable and widely used conservation intervention on islands, with demonstrated benefits for native species, ecosystems, and restoration outcomes (Jones et al., 2016). However, as eradications expand into larger, more complex, and increasingly inhabited systems (Oppel et al., 2011; Castaño et al., 2023), the transferability of previous experience declines and confidence in outcomes becomes less certain (Russell, 2025).

Across documented island rodent eradication attempts, approximately 88% have achieved eradication, although outcomes vary by method, species, and island context (Spatz et al., 2022). However, failures still occur, and in many cases the causes remain unclear despite identification of correlated risk factors (Holmes et al., 2015), requiring structured diagnosis to distinguish between multiple ecological and operational drivers (Kappes et al., 2019). Tropical and inhabited systems introduce biological and operational constraints (e.g. persistent rodent breeding, alternative foods, and access permissions) that increase the likelihood that some individuals are not exposed to the primary removal method (Keitt et al., 2015). Further, they often require co-development of operational strategies with local communities to ensure approaches are feasible and acceptable. Few large tropical island rodent eradications have been completed successfully, with several high-profile attempts failing despite reducing populations to low densities (e.g. Floreana, Henderson, Late). Thus, as project complexity increases the assumptions underlying standard eradication strategies become increasingly important.

Recent practitioner discussions emphasize that many high-risk eradications come close to success, with failure sometimes attributable to only a small number of surviving individuals and no clear explanation for how they persisted. In fact, several rodent eradications on inhabited Pacific islands (e.g. Lord Howe Island, Palmerston Atoll, Tetiaroa Atoll, and Wake Atoll) succeeded only after teams implemented targeted efforts to detect and remove remnant populations after the primary removal operation failed to eliminate the population (Harper et al., in press). These experiences suggest that success can depend as much on how projects respond to evidence of survivorship as on the initial population knockdown.

Given increasing project complexity and emerging evidence of survivorship on high-risk projects, practitioners face a planning challenge that is not fully addressed through conventional feasibility assessment. Most eradication planning processes appropriately focus on feasibility: whether legal permissions, social support, logistics, methods, funding, and operational capacity are sufficient to implement an eradication, and structured feasibility frameworks now provide standardized approaches to assess technical, ecological, and social readiness (Samaniego et al., 2026). Feasibility assessments can identify risks, dependencies, and knowledge gaps, but they do not necessarily resolve how survivorship should be incorporated into the implementation strategy when uncertainty remains and the consequences of failure are high. A project may be feasible in that a best-practice baiting operation can be delivered effectively, while still lacking a credible plan for detecting and responding to survivors. For some high-risk islands, the central planning question is therefore not whether eradication can be implemented, but whether the consequences of failure justify planning and resourcing for detection-triggered response. This paper presents a decision framework to help make that judgement more explicit during project planning.

2 A decision framework for strategy selection

Here, I present a practical decision tree for selecting an eradication strategy that accounts for survivorship risk, consequences of failure, and the capacity for early detection and response while eradication remains achievable. The framework distinguishes among three eradication strategy classes:

Single-event implementation: A strategy built around a time-limited operation that relies on one primary removal method supported by strong comparable precedent, followed by one-time eradication confirmation monitoring, with no predefined contingency response beyond limited mop-up.

Planned contingency response: A strategy in which the primary removal operation remains appropriate, followed by survivor-focused monitoring and predefined post-operation corrective actions designed to address specific, anticipated failure modes triggered by monitoring evidence.

Adaptive multi-method campaign: A phased strategy in which multiple removal methods are deployed strategically in sequence using decision thresholds based on evidence from survivor-focused monitoring because survivorship mechanisms remain unresolved or the consequences of failure justify greater response readiness.

The framework was developed for island eradications where complete removal is the objective and immigration can be prevented or managed. Some principles, including predefined decision thresholds and adaptive response, may also apply in mainland settings. However, mainland rodent management commonly involves continuing immigration and suppression objectives, which create different decision problems and fall outside the framework’s intended scope. Further, the framework distinguishes survivorship from reinvasion. It assumes that immigration has been prevented or reduced to a level consistent with eradication feasibility; where this condition cannot be met, sustained control or containment may be the appropriate objective.

Although many examples motivating this framework arise from single-event aerial toxicant operations, it is intended to apply irrespective of delivery method (e.g. toxicants, trapping, hunting, or biological control). The strategy classes instead describe how projects plan for uncertainty, survivorship, and response. In some contexts, particularly ground-based or inhabited-island eradications, adaptive campaign-style approaches may already represent standard practice. These strategy classes therefore represent common planning archetypes for managing survivorship risk for any island eradication.

The decision framework draws on insights from a workshop titled “Improving success rates for rodent eradication on high-risk islands” convened at the 2026 Island Invasives Conference in Auckland, New Zealand. The workshop brought together more than 100 practitioners, researchers, technical providers, policy makers, and funders to examine why some recent high-risk rodent eradications have narrowly failed despite close adherence to accepted best practice. Discussions focused on large, tropical, inhabited, and logistically constrained islands where environmental variability, alternative food, year-round breeding, bait competition, social constraints, and limited access increases uncertainty about the effectiveness of a toxicant-only based approach. Across presentations, plenary discussion, and breakout sessions, four recurring planning challenges emerged: determining whether evidence from past projects is truly transferable; understanding whether prior failures have been adequately explained; defining contingencies with clear triggers, tools, authority, and funding; and planning explicitly for survivorship. These themes were translated into a decision tree intended to make strategy selection more explicit during feasibility assessment, stakeholder engagement, peer review, and operational planning.

3 Key decision questions

The framework relies on a set of defined terms that distinguish among strategy classes, types of uncertainty, and monitoring approaches (Table 1). The decision tree (Figure 1) uses a sequence of six planning questions to guide projects into the three strategy classes. The distinguishing criteria, appropriate applications, practical requirements, and monitoring implications for each strategy class are summarized in Supplementary Table S1.

Table 1. Key terms and definitions for a decision tree for eradication strategy selection.

3.1 Define eradication success criteria and confidence requirements

Begin by defining what eradication success means in the local context and the level of confidence required before success can be declared. This includes the biological definition of success, the monitoring period, the evidence required, and the consequences of declaring success too early. Projects should also explicitly state their tolerance for failure risk, recognizing this may change as consequences or context evolve. Where tolerance for residual uncertainty is low, for example, in first-of-kind projects or those with high reputational or conservation consequences, a more robust strategy class may be required compared to projects designed to accommodate learning risk or staged progression.

3.2 Define comparable precedent

Then assess whether previous approaches on islands with similar socio-ecological characteristics provide a credible basis for confidence under local conditions. The central question is whether evidence from previous projects is sufficiently transferable to justify an expectation of success. For example, Harper (2020) argues that factors influencing the exposure of every rodent to the removal method on tropical islands may differ substantially from those that informed many temperate-island eradication strategies.

3.3 Identify poorly transferable or untested dependencies

If success depends on methods or assumptions lacking strong precedent in comparable contexts, projects would typically incorporate predefined detection and response pathways into the strategy. Where risks cannot be credibly bounded and corrected, projects would generally shift toward contingency or campaign strategies rather than only further optimization.

3.4 Examine prior unresolved evidence of survivorship

Where previous attempts on the same island or in comparable systems have failed, and the cause of survivors remains unresolved, repeating the same strategy without additional response capacity is unlikely to reduce risk (Samaniego et al., 2021). Prior failure would generally trigger explicit assessment of whether the failure mode is understood, whether a credible correction exists, and whether survivorship can be detected early enough to trigger a meaningful response. If the cause of prior failure remains uncertain, the framework would typically classify the project as an adaptive multi-method campaign rather than a single-event implementation with nominal contingencies. Put simply, confidence should be lower when an approach has previously failed under similar conditions and the reason for that failure remains unresolved, and thus the eradication strategy more conservative.

3.5 Evaluate consequences of failure

Projects with low consequences of failure may tolerate greater residual uncertainty, whereas projects with high consequences generally warrant stronger justification that risks are bounded, controlled, and/or mitigated. Consequences should be considered not only at the project level, but also in terms of broader organizational, regional, and sector-wide impacts. On some high-profile projects, failure may influence stakeholder confidence, future investment, regulatory support, and willingness to pursue similar interventions elsewhere. Where evidence is transferable and failure consequences are low, single-event implementation may be appropriate. Where failure would have high ecological, social, financial, regulatory, or credibility consequences, the burden of proof would be higher, and strategy selection would reflect whether risks can be credibly bounded rather than solely relying on further optimization. High consequences do not automatically require an adaptive multi-method campaign, but they do require a stronger justification for why single-event implementation or contingency response is sufficient.

3.6 Determine whether a predefined correction is credible

The final decision point distinguishes planned contingency response from an adaptive multi-method campaign. If anticipated risks can be clearly defined and addressed through predefined corrective actions, a contingency approach is appropriate. Where uncertainty remains material, failure modes are not well defined, or no single predefined correction is likely to resolve the risk, projects would typically be planned as an adaptive multi-method campaign. In these cases, survivor detection, response tools, governance triggers, and funding would generally be planned as core components of the eradication strategy rather than as optional backup. If no credible multi-method pathway can be defined, implementation would be deferred until feasibility, readiness, or strategy design is strengthened.

4 Application to high-risk eradications

Retrospective application of the decision framework to the Wake Atoll rodent eradication illustrates how the framework may be applied. A previous attempt on the 700-ha Pacific atoll removed one rat species but not the other. During the second attempt, survivors were detected following the initial removal operation, and an extended response phase was required to achieve eradication. During initial design for the second attempt, comparable precedent was limited, prior eradication failure was present, and the mechanisms of survivorship remained unresolved (Kappes et al., 2019). Consequences of failure were high due to the site’s conservation significance and the implications of a second unsuccessful attempt. Under these conditions the framework would have favored planning the project as a multi-method campaign. In practice, the initial single-event implementation was followed by an extended response phase once survivors were detected. This example illustrates a central premise of the framework: where survivorship cannot be confidently ruled out and the consequences of failure are high, success may depend as much on planning and resourcing for survivor detection and response as on the primary removal operation itself.

5 Discussion

5.1 Why strategy selection matters

The consequences of rodent eradication failure are not distributed equally across projects. On many islands, a single-event implementation remains an appropriate and effective strategy. However, for some high-risk projects, a small number of surviving individuals can determine the outcome, and the consequences of failure may extend well beyond the project itself. In these contexts, success may depend less on assuming survivorship is unlikely and more on making explicit decisions about how it will be managed if it occurs.

Feasibility assessments play a critical role in shaping project design (Samaniego et al., 2026). The challenge is less in recognizing survivorship as a possibility than in deciding how much weight it should carry in strategy design when the stakes are high. In practice, projects can acknowledge the possibility of survivors yet still proceed as single-event implementations because precedent is judged sufficient to justify that choice. Workshop discussions highlighted that this assumption becomes increasingly difficult to justify as ecological, operational, and social complexity increase. Adaptive, phased, and campaign-style approaches have long been used in some rodent eradication contexts, particularly ground-based and inhabited-island operations where repeated detection and response are already integrated into implementation (e.g. Bell, 2019).

Making decisions about survivorship more explicit during project planning can help practitioners determine when confidence in a primary removal operation is sufficient and when the consequences of failure justify a more adaptive approach. In doing so, it may also support the responsible evaluation of innovative methods, strengthen the evidence base for future projects, and improve confidence in when more adaptive approaches are warranted.

5.2 Clarifying contingency versus adaptive campaign approaches

Not all contingencies provide meaningful risk reduction. Their value depends on whether failure modes are understood, survivorship can be detected early enough to trigger a predefined response while success remains achievable, and the authority and funding to act are already in place. Without these elements in place, contingency plans may provide little practical value and can leave important risks unresolved until there is evidence of survivorship. This distinction reinforces the role of adaptive management in high-risk eradications. As discussed by workshop participants, adaptive management is not simply the ability to adjust tactics after implementation, but a structured, pre-planned process that establishes decision points, response options, milestones and learning loops in advance. Where uncertainty about survivorship remains consequential, adaptive management is most effective when it is built into the eradication strategy, budget, and decision-making process early in project design and co-designed with project stakeholders.

5.3 Implications for practitioners and stakeholders

For practitioners, local stakeholders, and funders, the framework provides a transparent way to evaluate strategy selection, evidence transferability, and response readiness. For funders, it clarifies when and why some eradications require additional investment by distinguishing projects that can be responsibly financed as single-event implementations from those that may warrant contingency responses or adaptive multi-method campaigns with explicit detection and predefined response capacity built in.

More broadly, the framework implies a shift in how some high-risk rodent eradications are assessed, planned, and funded. On some high-risk islands, success may depend on planning for a phased eradication effort from the outset, with detection and response treated as integral components of implementation, similar to approaches commonly used in pig and cat eradications. This has direct implications for funder communication and project design. Teams would be prepared to explain why certain projects require larger initial budgets, longer timelines, and flexible or staged funding to increase the likelihood of removing survivors before failure becomes entrenched. Although this may raise upfront costs, it can substantially reduce the risk of total failure and avoid the far greater cost, delayed realization of conservation and social benefits, and erosion of stakeholder confidence that may result when high-profile projects require repeated attempts. Alternative funding approaches may therefore be worth exploring for high-risk projects.

A related challenge is how to plan and resource contingent implementation phases in practice. Adaptive and campaign-style approaches introduce uncertainty not only in cost, but in the timing, scale, and type of response actions required. This creates operational challenges for staffing, logistics, vendor contracting, and risk management, particularly where response capacity may be needed but not ultimately used. While this framework helps identify when adaptive approaches may be warranted, practical implementation remains a significant challenge. Developing workable models, particularly for large or remote islands, remains an important area for future collaboration and innovation.

Sustaining success rates on increasingly complex islands will also require stronger shared learning systems. Regular peer review, post-operation debriefs, and globally accessible repositories of operational experience can help reduce repeated failures, improve interpretation of uncertain outcomes, and accelerate learning across projects, consistent with calls to disseminate results and lessons learned more widely (Bird et al., 2019).

5.4 Limitations and future use

The framework is intended to support structured judgement rather than predictive certainty (Russell, 2025). The framework is qualitative and does not estimate eradication probability, rank projects by risk, or prescribe specific tools or methods. Because the framework relies on practitioner judgement, it remains vulnerable to overconfidence and misclassification, particularly where important ecological, operational, or social differences remain unresolved. Independent peer review of strategy classification should generally be considered standard practice for high-risk projects.

Important uncertainties remain regarding rodent behavior, bait uptake, detection probability, population dynamics, environmental influences on bait effectiveness, and interpretation of survivor evidence. These uncertainties influence how confidently survivorship can be treated as bounded and whether more adaptive strategies may be warranted.

Future use of the framework should therefore focus on testing and refinement in real planning contexts. Retrospective application to completed eradications could assess whether projects that failed or required substantial follow-up would have been classified differently during planning. Prospective application during feasibility assessment could test whether explicit strategy classification improves assumption testing, monitoring design, contingency budgeting, governance, and response readiness. Continued refinement will also depend on stronger shared learning systems and regular peer review. These mechanisms can improve interpretation of uncertain outcomes, reduce repeated failure, and support more consistent strategy selection across projects.

6 Conclusion

Rodent eradication best practices remain highly effective and continue to deliver conservation and social benefits across a wide range of island contexts. However, on some high-risk islands the key planning question is whether the primary removal operation is likely to be sufficient, or whether projects should anticipate survivorship and plan accordingly. The framework presented here is intended to make that judgement more explicit by helping practitioners evaluate precedent, unresolved evidence of survivorship, consequences of failure, and response readiness. As eradications move into increasingly complex ecological and social contexts, success may depend as much on planning for how survivorship will be managed as on optimizing the primary removal operation. Used as intended, the framework helps retain the efficiency of traditional approaches where appropriate while encouraging more deliberate planning for detection-triggered response when the stakes of failure are high.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. During the preparation of this manuscript, the author used Microsoft Copilot 365 for AI-assisted writing support, including editing for clarity and readability and assistance with articulating conceptual arguments. All AI-assisted content was critically reviewed, revised, and integrated by the author, who takes full responsibility for the final manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcosc.2026.1957591/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Francesco Maria Angelici, National Center for Wildlife, Saudi Arabia

Reviewed by: Adedayo Michael Awoniyi, Federal University of Bahia, Brazil