Half of resources in threatened species conservation plans are allocated to research and monitoring

Threatened species recovery occurs under severe budget constraints, creating a central resource allocation problem: how to balance spending between direct conservation actions and research and monitoring (RM) that informs those actions. Insufficient information can lead to ineffective or counterproductive management, but excessive or poorly targeted RM can divert scarce resources from action. Prior work has highlighted that funding is far below needs (e.g., U.S. expenditures are roughly 25% of what recovery plans require) and that managers must weigh costs and benefits of actions against the value of information. RM, defined here as activities generating information on species, threats, socio-economic contexts, responses to interventions, and effectiveness of techniques, can improve decisions when linked to action (e.g., adaptive or state-dependent management). However, non-strategic RM can waste resources, delay decisions, and sometimes monitors populations without plans for action, with some species monitored until extinction. While prior studies have optimized threatened species recovery as a cost–benefit allocation problem, recovery is unlikely if most resources are allocated to RM without clear decision links. This study asks: what proportion of recovery budgets is allocated to RM across jurisdictions and taxa; which characteristics are associated with higher RM allocations; and how do RM allocations relate to recovery outcomes? The authors analyze recovery plans for over 2300 species from the U.S., New Zealand (NZ), and New South Wales (NSW, Australia) and provide recommendations to ensure RM informs action and improves recovery efficiency.

The paper builds on literature documenting funding shortfalls for biodiversity conservation and the need for evidence-based, adaptive management. Prior studies critique monitoring that is not linked to decision-making, noting cases of surveillance without action and instances where species were monitored to extinction. Research has examined endangered species recovery as a resource allocation problem and highlighted mismatches between proposed and realized budgets and weak links between funding and outcomes due to data quality. Decision tools such as cost-effectiveness analysis and Value of Information (VOI) have been proposed to optimize trade-offs between action and information. The authors also contrast RM allocations with R&D spending in other sectors to illustrate the comparatively high share of conservation budgets devoted to information gathering.

Study scope and data sources: The authors compiled recovery planning data for 2328 species/subspecies/distinct populations across three jurisdictions: NZ (700), NSW (361), and the U.S. (1267 terrestrial and freshwater taxa). These represent the most threatened listed species and/or those with recovery plans. Recovery planning databases provided proposed management tasks, their descriptions, and cost estimates over a 50-year horizon. For NZ and NSW, tasks were developed via structured expert elicitation within systematic prioritization frameworks; for the U.S., tasks and costs were extracted from published recovery plans. Classification of tasks: Each management task was categorized as research and monitoring (RM) or action using IUCN classification scheme definitions. For NZ/NSW, keyword searches (e.g., survey, monitor, research, inventory) guided identification, followed by manual review. For the U.S., tasks were classified manually by the first author and a trained technician; the first 200 tasks were double-coded to ensure consistency. Ambiguous tasks were discussed to reach consensus. Tasks with vague descriptions were excluded (2.6% of U.S. tasks). Some tasks (3.9%) were split as both action and RM (e.g., translocation with post-release monitoring); in such cases, costs were apportioned using the average proportional difference between action and RM for each jurisdiction. A subset of 8050 U.S. tasks (207 species) was further subcategorized into 17 RM types to summarize common activities. Cost calculations and RM proportion: Costs for each task were standardized over 50 years. For each species, the proportion of the proposed budget allocated to RM was computed as the total cost of RM tasks divided by the total cost of all tasks. Species with zero proposed budgets (23 U.S. species) and extinct species were removed from analyses. Covariates: Cross-jurisdictional covariates included taxon (amphibians, birds, bryophytes, fishes, fungus, invertebrates, mammals, reptiles, vascular plants), total proposed budget, and estimated benefit of implementing all management tasks (expert-elicited in NZ/NSW; approximated via Recovery Priority Numbers in the U.S.). Additional U.S.-specific variables examined included listing status, proportion of RM tasks completed, first fiscal year of earliest RM, number of species per plan, proportion of RM tasks assigned high priority, and a derived recovery potential metric from RPN. Statistical analysis: Beta regression models (R betareg) related the proportion of budget allocated to RM to species/jurisdictional covariates. One model included all jurisdictions; a second used the expanded U.S. covariate set. Continuous covariates were standardized; categorical covariates were dummy-coded with careful reference category selection to minimize multicollinearity (VIF < 2 in final models). Extremely large-budget outliers (five species > US$5M; results robust to inclusion/exclusion) were removed to improve fit. Recovery outcomes: Recovery indices were compiled from prior U.S. biennial reports to Congress (1989–2011; scores −11 to +11) and constructed analogously for NZ (4 assessments) and NSW (5 annual report cards) using −1/0/+1 for decline/stable/improve between assessments. This yielded recovery indices for 78.5% (U.S.), 13.5% (NZ), and 14.7% (NSW) of species. Data/code availability: Datasets and analysis code are available via Figshare; recovery status data were sourced from jurisdictional reporting systems.

• Overall allocation: On average, 50% ± 27% (sd) of proposed recovery plan budgets are allocated to research and monitoring (RM) across 2261 species with nonzero budgets in the U.S., NZ, and NSW. About 4% of species (3% U.S., 6% NZ, 2% NSW) had >95% of proposed budgets allocated to RM.
• Jurisdictional differences: Mean RM proportions were higher in the U.S. (52% ± 24%) and NZ (52% ± 28%) than in NSW (36% ± 28%). Older U.S. plans allocated more to RM than newer plans, indicating a temporal decline in RM allocation in recent planning.
• Taxonomic patterns: Bryophytes (NZ only) had the highest RM budget shares overall; in the U.S. and NSW, amphibians had the highest RM shares. Birds consistently had the lowest RM shares across jurisdictions.
• Budget and benefit effects: Species with larger total proposed budgets allocated lower proportions to RM; small-budget species showed wide variability (0–100%). Higher estimated benefits of implementing all management actions were associated with lower RM shares (weaker trend in the U.S., likely due to how benefit was estimated from RPN).
• U.S.-specific drivers: A higher proportion of RM tasks assigned high priority and earlier initiation of RM (older first RM fiscal year) were associated with higher overall RM budget shares.
• Recovery outcomes: Species with higher RM budget shares had poorer recovery indices. In the U.S., species with recovery indices 9–11 (declining in 9–11 of 11 reports) had a median of 70% of budgets allocated to RM. In NZ and NSW, species with indices −2 to −3 (declining in most assessments) had median RM shares of 44%.
• Trend over time and examples: RM allocation appears to be decreasing in newer plans, aligning with a shift from surveillance to targeted adaptive monitoring. The Florida Scrub-Jay provides an illustrative case: a 1990 plan emphasized RM, whereas a recent draft allocates <1% to ongoing RM and prioritizes habitat protection based on prior research insights.

The study shows that allocating roughly half of threatened species’ proposed recovery budgets to RM is common, and higher RM shares are associated with poorer recovery outcomes. This does not imply RM is inherently detrimental; rather, recovery is unlikely if plans prioritize information gathering without clear decision rules linking information to action. Several mechanisms may underpin the observed relationship: (1) plans heavily weighted to RM may delay necessary threat abatement; (2) greater uncertainty and fear of adverse outcomes for highly imperiled species may bias planners toward RM over action; and (3) species with limited baseline knowledge may demand more RM. Despite these factors, deliberate rebalancing toward actions, or RM that directly informs imminent decisions, could improve outcomes. Jurisdictional policy differences matter: NSW’s lower RM allocations reflect guidance to include RM only when it informs specific actions, and both NZ and NSW increasingly prioritize RM for species with uncertain trends and threats, while emphasizing action where declines are understood. Such policy structures likely foster a more effective RM–action balance. Decision-support tools can guide optimal allocations, including cost-effectiveness analyses and Value of Information (VOI) to quantify when information will improve decisions enough to justify its cost. Sensitivity analyses within prioritization frameworks can identify key uncertainties driving decisions. Considering socio-economic context is critical, as this study focuses on relatively resource-rich jurisdictions. Overall, carefully limiting RM to efforts that directly enhance the ability to act can preserve resources for implementation, potentially improving recovery outcomes.

This work provides the most comprehensive cross-jurisdictional assessment to date of how recovery plan budgets are divided between research/monitoring and action. It reveals that, on average, half of proposed budgets are assigned to RM, with substantial variation among jurisdictions and taxa, and that higher RM allocations are associated with poorer recovery indices. The findings underscore the need to design RM that explicitly informs decisions and to avoid defaulting to information gathering when action is warranted. Future directions include: applying VOI and cost-effectiveness analyses to determine optimal RM–action splits for specific species and contexts; embedding explicit decision rules and triggers in monitoring plans; evaluating how realized (not just proposed) spending sequences influence RM–action shares; improving the quality and coverage of recovery status data; and extending analyses to less resource-rich regions to understand context-dependent allocations and outcomes.

• Proposed vs. realized actions: Analyses use proposed recovery plan budgets and tasks; actual implementation often falls short and may occur in sequences that inflate the realized RM share beyond proposed values.
• Coverage and data quality: Recovery indices were available for subsets of species (U.S. 78.5%, NZ 13.5%, NSW 14.7%) and prior studies note limitations of recovery assessments. This constrains inference about outcomes.
• Classification challenges: Distinguishing RM from action can be ambiguous for some tasks; 3.9% of tasks were split between RM and action, and 2.6% of vague U.S. tasks were excluded. Some categorizations required judgement and consensus.
• Benefit estimation (U.S.): The proxy for benefit derived from Recovery Priority Numbers has recognized limitations and may weaken estimated relationships.
• Generalizability: Results are from three, relatively resource-rich jurisdictions and may not generalize to all governance or funding contexts.
• Outliers and model choices: Extremely high-budget species were excluded to improve model fit (though results were robust), and multicollinearity led to exclusion of some correlated covariates in final models.