Geology topics

Research about Guam

Source-linked reports with geographic coverage including Guam.

At least 19 recordsLinked to original sources

USGS2026

Demographic responses to experimental manipulation of an enclosed population of invasive brown treesnakes (Boiga irregularis)

Understanding how invasive species populations respond to perturbations can give insight into effective management. However, opportunities to test theoretical population responses are rare, as experimental perturbations often run counter to management goals (e.g., suppression). On the island of Guam, a 5-ha outdoor enclosure (with no immigration or emigration) constructed in 2003 serves as a unique population-scale study area for invasive brown treesnakes ( Boiga irregularis ). Capture-mark-recapture studies (CMR) and several experiments have taken place since 2004. Here, we analyze a subset of the CMR data collected between 2008 and 2012, a time period encompassing two experimental perturbations: (1) supplemental feeding to determine if the population is food limited and (2) targeted removal of adult and large-subadult cohorts to quantify the timing and magnitude of the population's demographic response to a simulated control effort. We used the CMR data to fit a Pradel model in a Bayesian framework to compare the population growth rate, survival probability, and recruitment rate prior to, during, and post perturbations. We found that increasing food resources increased population growth and recruitment rates, but that survival remained relatively constant. The removal perturbation resulted in a decrease in population growth rate and survival immediately after the removal, but population growth rate and recruitment began increasing 1 year after the removal perturbation. Our results suggest that brown treesnakes in the enclosed population are prey limited, likely constraining population growth, and that the suppression of population growth after removal of adults and large sub-adults is short-lived (i.e., less than a year).

Guam↗

USGS2025

A comparative evaluation of bait types taken by brown treesnakes or other nontarget species, including ants

Evaluating invasive species removal can help improve efficacy or application of control tools. Toxic baiting is a primary method used to suppress many invasive species, including the brown treesnake ( Boiga irregularis ) in Guam. Based on limited alternative approaches to assess tool efficacy in snake removal, bait take is often used as a proxy for snake removal, despite evidence that nontarget species remove baits in a spatially heterogeneous manner. Using photographic validation, we tested take rates of mouse, chick, and rat baits by brown treesnakes and nontargets, including invertebrates, at three sites undergoing snake removal for vertebrate conservation goals. We also identified the ant species present on baits for a subset of baits. Take rates by snakes and nontargets varied by site. We found that nontarget take (22–42%) was considerably greater than take by brown treesnakes (1–13%). In addition, baits that had ants were taken less often by snakes as compared to baits without ants (7–12% vs. 8–52%, respectively). The dominant ant species collected from ant bait stations affiliated with baits was Solenopsis spp. Ant presence appeared to have deterred brown treesnakes from consuming baits, and ant removal or prevention may improve take rates by brown treesnakes. There was not, however, a consistent correction factor that could be applied across sites to estimate brown treesnake take rates of baits without photographic validation.

Pacific Science↗

USGS2025

Using periodic matrix models to simulate the effectiveness of alternative reintroduction strategies for lizards on a seasonal tropical island

Conservation translocations and reintroductions are widely used to improve conservation outcomes for declining species. Reintroductions are unlikely to be successful if the threats that led to the extirpation of the focal species, such as non-native predators, have not been ameliorated. The non-native brown treesnake ( Boiga irregularis ) was introduced to Guam in the mid-20th century and has devastated the native lizard fauna of this Pacific Island. Native lizard populations persist on small islands near Guam and could act as sources for reintroductions. Recently, trapping methods have been shown to greatly reduce the density of brown treesnakes, raising the possibility that native lizards could be reintroduced where brown treesnake abundance is controlled. Here we developed demographic models for four lizard species present on Guam to assess their validity as a trial for reintroducing species to meet native species recovery goals. These four species vary in their size, activity, and susceptibility to predation by brown treesnakes. We used periodic matrix models to simulate changes in lizard vital rates driven by the wet-dry seasonal cycle found on Guam. We found that the release of 60 individuals of each species was likely to result in successful reintroduction outcomes provided adult survival in the wild is similar to rates estimated based on life history parameters. Demographic models can be used to forecast reintroduction outcomes, including predicting the probability of success and evaluating causes of failure in the event that populations do not establish.

Guam↗

USGS2025

Reproductive condition of an invasive snake in urban, savanna, and forest habitats

Predation by Brown Treesnakes ( Boiga irregularis ) has caused the decline or extinction of all native forest vertebrates on Guam, including birds, lizards, and bats. Loss of the highest-quality prey has caused shifts in Brown Treesnake size and life history. We sought to understand how reproductive condition varies among habitats island-wide, particularly comparing urban and savanna habitats (where some primarily non-native large prey remain) with prey-depauperate forest habitats. We dissected 100 snakes from each of 18 sites stratified by 6 habitat types and calculated a body condition index, relative follicle size (for females; N = 817) and relative testes size (for males; N = 992) for each. Low detection of reproductive females continues to hamper research on their biology, as only 11% of females had secondary vitellogenic follicles (N = 89). Although variation among sites within habitats was high, above-average reproductive development of both males and females was most frequently observed in urban and savanna habitats, emphasizing the potential for prey subsidies in these habitats to fuel population growth and affect invasive snake management objectives. Furthermore, the trade-off between body condition and reproductive condition appeared to be more severe in forested habitats and less severe in urban habitats for both males and females. There were few obvious seasonal patterns, although there was a slightly lower probability of females having secondary vitellogenic follicles in the wet season compared to the dry season.

NeoBiota↗

USGS2025

Growth rate variation in Brown Treesnakes (Boiga irregularis): An invasive species of conservation concern

Somatic growth rate is a fundamental trait that influences metabolism, lifespan and reproductive maturity and is critical for understanding population dynamics and informing management actions. Brown Treesnakes ( Boiga irregularis ) introduced to Guam are highly invasive and can reproduce year-round without discrete cohorts. We compared snake size trajectories described by the conventionally used von Bertalanffy growth function versus the Gompertz model. Using quantile regression with a regularized effect for individual snakes we modeled growth rates of 270 marked, wild snakes as a function of size. The Gompertz model explained more of the variation in growth and rendered more realistic predictions of asymptotic sizes than did the von Bertalanffy model. With the Gompertz model, growth rates were 1.05–1.16× faster in males than in females. Females reached asymptotic sizes at shorter snout-vent lengths than males. Growth rate was positively correlated with amount of precipitation, and modeling wet-dry seasonality on Guam as a sinusoidal function identified a growth peak in September—October. Effects of seasonality and precipitation, however, were minor compared to individual and sex related differences in size-adjusted growth rates. We estimated that the 50th (and 5th, 95th) growth-rate percentile males in our study population become sexually mature at an age of 33 (∞, 15) months, while females mature at 41 (∞, 18) months, where ∞ indicates that the slowest growing snakes never reach maturity. However, 50% of the snakes mature at a size below the median, and age at maturity may be as low as 10.4 (males) and 13.7 (females) months for average-sized hatchlings that grow fast. Our results have implications for the timing of management options for this species and our approach can be broadly applied to animals where repeated growth data are obtained and age is unknown.

Ecology and Evolution↗

USGS2025

Automated methods for processing camera trap video data for distance sampling

Context Population monitoring is an essential need for tracking biodiversity and judging efficacy of conservation management actions, both globally and in the Pacific. However, population monitoring efforts are often temporally inconsistent and limited to small scales. Motion-activated cameras (‘camera traps’) offer a way to cost-effectively monitor populations, but they also generate large amounts of data that are time intensive to process. Aims To develop an automated pipeline for processing videos of ungulates (Philippine deer, Rusa marianna ; and pigs, Sus scrofa ) on Andersen Air Force Base in Guam. Methods We processed camera videos with a machine learning model for object detection and classification. To estimate density using distance sampling methods, we used a separate machine learning model to estimate the distance of target animals from the camera. We compared density estimates generated using manual versus automated methods and assessed accuracy and processing time saved. Key results The object detection and classification model achieved an overall accuracy >80% and F1 score ≥0.9 and saved 36.9 h of processing time. The automated distance estimation was fairly accurate, with a 1.1 m (±1.4 m) difference from manual distance estimates, and saved 16.8 h of processing time. Density estimates did not differ substantially between manual and automated distance estimation. Conclusions Machine learning models accurately processed camera videos, allowing efficient estimates of density from camera data. Implications Further adoption of motion-activated cameras coupled with automated processing could lead to continuous, large-scale monitoring of populations, helping to understand and address changes in biodiversity.

Guam↗

USGS2025

Using distance sampling with camera traps to estimate densities of ungulates on tropical oceanic islands

Reliable population estimates are one of the most elementary needs for the management of wildlife, particularly for introduced ungulates on oceanic islands. We aimed to produce accurate and precise density estimates of Philippine deer ( Rusa marianna ) and wild pigs ( Sus scrofa ) on Guam using motion-triggered cameras combined with distance sampling to estimate densities from observations of unmarked animals while accounting for imperfect detection. We used an automated digital data processing pipeline for species recognition and to estimate the distance to detected species. Our density estimates were slightly lower than published estimates, consistent with management to reduce populations. We estimated the number of camera traps needed to obtain a 0.1 coefficient of variation was substantial, requiring > ten-fold increase in camera traps, while estimates with precision of 0.2 or 0.3 were more achievable, requiring doubling to quadrupling the number of camera traps. We provide best practices for establishing and conducting distance sampling with camera trap surveys for density estimation based on lessons learned during this study. Future studies should consider distance sampling with camera traps to efficiently survey and monitor unmarked animals, particularly medium-sized ungulates, in tropical, oceanic island ecosystems.

Journal of Tropical Ecology↗

USGS2025

The accuracy of capture per unit effort in predicting density of a cryptic snake was more sensitive to reductions in spatial than temporal coverage

A critical component of monitoring wildlife populations is understanding changes in population size or abundance. However, for most populations a complete census is not possible; thus, trends or abundance need to be estimated through alternative means, such as indexes. An important aspect of using indexes, such as capture per unit effort (CPUE), is validating them as accurate or precise predictors of population trends or abundance. We completed such analyses using data collected from visual surveys and trapping for brown treesnakes ( Boiga irregularis ) within a 5-ha enclosure that was undergoing a continuous population decline. During a ~ 6-year period, we censused and marked the snake population to fully enumerated the population, with new individuals resulting from births and removals resulting only from mortality (natural or experimental). From trapping and visual surveys, we were also able to calculate CPUE as a function of trap nights or km surveyed and used regressions to forecast snake density (snakes/km) in the enclosure from CPUE. We also rarefied the true dataset to measure whether reductions in sampling intensity, either temporally or spatially, affected the accuracy or precision in predicting snake density from CPUE. We found that trap CPUE demonstrated no statistical relationship to density based on our study methods. CPUE during visual surveys did predict actual density, with sufficient spatial and temporal sampling intensity. CPUE from visual surveys was relatively robust against reductions in temporal sampling when spatial intensity remained high. However, reductions in the spatial area covered to less than 50% of the enclosure rapidly reduced the accuracy and precision in using CPUE to forecast density. Our results indicate that visual surveys are a relatively accurate measure of true density for brown treesnakes, given sufficient spatial sampling effort. The spatial area of coverage required for CPUE to accurately predict changes in abundance was, however, intense with > 50% of the spatial area required to be sampled on a given sampling night. Our results indicate that CPUE is only reliable as an index of abundance or population trends for cryptic snakes, if sampling effort covers most of the landscape over which populations are being estimated.

Guam↗

USGS2025

Development and evaluation of the remote passive integrated transponder tag reader for customizable monitoring of wildlife

Increasing the quantity and quality of mark-reencounter data can be important when individual identification data are needed to address research or management questions. Physical recapture methods tend to be labor-intensive and therefore expensive, while remote detection methods are not easily applied to all taxa or are proprietary and thus difficult to customize. However, partnerships between wildlife scientists and engineers have the potential to provide innovative solutions that improve data collection while reducing costs. We describe a collaborative effort to improve the collection of mark-reencounter data on an invasive reptile, the brown treesnake ( Boiga irregularis ). In laboratory trials conducted on Guåhan (CHamoru; Guam in English) in 2021, we assessed the feasibility of detecting snakes fitted with passive integrated transponder (PIT) tags using customized remote PIT-tag readers (RePTaR) paired with baited snake traps. We evaluated differences in scanning success as a function of snake traits, location of PIT tags within snakes, and the distance of the PIT tag from the reader. We successfully scanned all individuals, on average 529 times but ranging from 3 to 6,436 times, during 12-hour trials using a designed customizable RePTaR unit. Snake characteristics and marking location explained little of the variability in scanning success, with distance from the reader as the best-supported covariate. By building cross-disciplinary partnerships, we can develop customizable technological solutions and provide additional tools with which to address challenges in wildlife research and management.

Guam↗

USGS2024

Quantifying the importance of ontogeny and prey type in modeling top-down and bottom-up effects of an ectothermic predator

Dietary decisions by predators can affect prey abundance and overall food web dynamics. Many predators do not forage on the same prey at the same frequency throughout their lives. Ontogenetic shifts in prey preference are not, however, often accounted for when modeling food web relationships, despite growing literature that suggests that stage specific dietary relationships may be an important consideration when modeling trophic interactions. We investigated the importance of considering size-structure of a predator population with ontogenetic diet shifts in evaluating relationships with prey response using a manipulative experiment with the brown treesnake ( Boiga irregularis ) in Guam. After removing ~ 40% of the snake population via toxic mammal carrion, we measured the strength of the relationship between snake density and the response of two types of prey (lizards and mammals). We evaluated these relationships based on total population size or division of the population into stage specific size categories based on diet preference predictions. We hypothesized that the density of juvenile snakes would correlate more strongly with lizard detections, while adult snakes would better correlate to rodent detections. We also measured reproductive output following changes in rates of prey detection. As expected by known ontogenetic shifts in dietary preference, explicit stage-based models better predicted shifts in rates of observed prey items than did total predator density for both lizards and mammals. Additionally, rodent detections were predictive of one reproductive pulse from snakes, while lizard detections were not predictive or correlated. Our findings support that consideration of predatory species stage-based dietary preference can be meaningful for understanding food web dynamics, particularly when a predator has a broad diet or one that changes through time.

Guam↗

USGS2024

Gape-limited invasive predator frequently kills avian prey that are too large to swallow

Gape-limited predators (e.g., snakes, many fish) are not generally expected to pose a predation threat to prey that are too large for them to swallow. However, the extent to which snakes predate on prey that exceed their gape limitation remains largely unknown. We conducted the first study to investigate the influence of both prey and predator sizes on the frequency of ingestion success by snakes in a natural system. We combined survival monitoring of an avian prey species ( Aplonis opaca ) via radio-telemetry with a survey of the size distribution of their major predator ( Boiga irregularis ) on Guam. This allowed us to assess (1) the frequency of unsuccessful ingestion by the predator, (2) whether the size of the prey predicts ingestion success, (3) whether the size of the predator predicts ingestion success, and (4) the relationship between prey and predator sizes in successful ingestion attempts. We found that nearly half (47.95%) of ingestion attempts by snakes on fledgling birds were unsuccessful, and no instances where unsuccessful ingestion caused the mortality of the snake. Attempts to consume smaller fledglings were as likely to be unsuccessful as attempts to swallow larger fledglings. However, snakes that successfully ingested fledglings were among the largest snakes in the population, and larger than average conspecifics attracted to endothermic prey. The smallest snakes that successfully ingested fledglings attained remarkably high relative prey mass values for their species, consuming prey weighing up to 79.9% of their own mass. Our study indicates that B. irregularis routinely predate prey that are too large for them to successfully ingest, which causes mortality to the prey but poses little risk to the predator. The potential reward for snakes in consuming oversized prey may outweigh the inherent risks, while instances of predation that do not result in consumption may have considerable impacts on prey populations.

Guam↗

USGS2024

A review of the ecology and conservation biology of Sali (Micronesian Starling, Aplonis opaca guami) on Guam

The accidental introduction of the Brown Treesnake (BTS, Boiga irregularis ) to Guam following World War II led to the extinction, extirpation, or severe decline of most of Guam’s native avifauna. One forest bird species that managed to persist is the cavity-nesting Såli (Micronesian Starling, Aplonis opaca guami ), a once-ubiquitous native omnivore whose current Guam distribution is restricted primarily to urbanized areas in the northern half of the island, with a second much smaller population on Cocos Island off southern Guam. Såli on Guam breed year-round and are habitat generalists with large home ranges. Besides arthropods and small vertebrates, they consume fruit from a wide range of native and introduced plant species and are key seed dispersers within the local ecosystem. Såli on Guam suffer extremely high post-fledging mortality due to predation by BTS and cats, leading to a population age structure strongly skewed towards adult birds. Other threats include habitat loss and degradation, extreme weather events, and sea level rise. Key management activities aimed at restoring this species include predator control, while non-native ungulate removal and vegetation management may be undertaken to support broader ecological recovery. This species account on Såli provides a review of the literature on this species, synthesizing information on habitat use, life history, and demographics to provide a biological foundation for future recovery efforts.

Micronesica↗

USGS2024

Videographic monitoring at caves to estimate population size of the endangered yǻyaguak (Mariana swiftlet) on Guam

The yǻyaguak (Mariana swiftlet; Aerodramus bartschi ) is an endangered cave-nesting species historically found on Guam and the southern Mariana Islands, Micronesia. The population on Guam has been severely affected by the introduction of the brown treesnake Boiga irregularis . Population status assessments have, however, been challenging due to the limitations of traditional counting methods, which rely on visual observations at cave entrances and are prone to inaccuracies. To improve count accuracy, we estimated yǻyaguak population size and relative nesting activity using thermal and near-infrared videography. The population on Guam was surveyed at the island’s 3 known occupied caves (Mahlac, Maemong, and Fachi) between 2019 and 2023. Mahlac Cave harbored the largest colony, which ranged from 506 to 665 birds; Maemong Cave held 144 to 196 birds; and Fachi Cave, which is sometimes flooded, had 28 (in 2019) and 35 birds (in 2023). Our estimates indicate a slight decline in the yǻyaguak population over the study period. This study demonstrates the potential of thermal and near-infrared videography for improved monitoring of yǻyaguak colonies and nesting activity, which will contribute to our understanding of population dynamics and the effectiveness of management strategies such as brown treesnake control.

Guam↗

USGS2024

Post-Typhoon Mawar population counts of the endangered yǻyaguak (Mariana swiftlet) on Guam

The yǻyaguak (Mariana swiftlet, Aerodramus bartschi) is an endangered cave-roosting species native to Guam and southern Mariana Islands, Micronesia. The population on Guam has declined substantially over the last half century, likely due to the introduction of the brown treesnake (Boiga irregularis), but other factors have been proposed including habitat loss, pesticides, reduced food resources, and powerful storms. On May 24–25, 2023, Guam was hit by the category-4 Typhoon Mawar, which brought wide-spread destruction to the island. Approximately 6 weeks prior to the typhoon in April 2023, we conducted surveys of the three known colonies of yǻyaguak at Mahlac, Maemong, and Fachi caves. This survey provided a baseline population estimate prior to Typhoon Mawar’s landfall. In July 2023, we resurveyed the caves to estimate colony size and assess whether the cave colonies experienced significant declines in the immediate aftermath of the storm. In November 2023, we conducted our regularly scheduled biannual surveys which provided a longer-term assessment of potential impacts from the typhoon. Our counts indicated that in the immediate aftermath (~6 weeks) there was a 7–8% reduction in size of colonies compared to the April 2023 counts, but in November there were larger, 19–35% reductions in the cave colonies from pre-Mawar surveys. These results suggest that the long-term effects of the storm were more detrimental to the yǻyaguak populations than the immediate impacts if these colony size reductions are a result of the storm. Continued monitoring of the yǻyaguak cave colonies will allow for documenting the pace of recovery from the storm.

Guam↗

USGS2023

Suppression of invasive Brown Treesnakes and reintroduction of native avifauna on Guam

Many agencies and interest groups are committed to re-establishing components of Guam's native avifauna through the reintroduction of captive-reared birds or translocation from other islands in the Marianas if the Brown Treesnake ( Boiga irregularis ; BTS) can be eliminated. Island-wide eradication of BTS from Guam continues to appear out of reach, but with recent and future advancement in BTS suppression technology, we may soon reach a point where suppression may enable the reintroduction of some of Guam's extirpated bird species. Our simulations indicate that bird persistence in Guam's forests would only be possible at a suppression level just short of full eradication.

Bulletin of the Ecological Society of America↗

USGS2023

Modeling non-structural strategies to reduce pedestrian evacuation times for mitigating local tsunami threats in Guam

Reducing the potential for loss of life from local tsunamis is challenging for emergency managers given the need for self-protective behavior of at-risk individuals within brief windows of time to evacuate. There has been considerable attention paid to discussing the use of tsunami vertical-evacuation structures for areas where there may be insufficient time to evacuate. This strategy may not be feasible or needed for at-risk populations in island communities for multiple reasons. We examine the influence of three non-structural interventions (reducing departure delays, increasing travel speeds, and managing vegetation to create new paths) that may improve the evacuation potential for at-risk individuals in island communities and use the United States territory of Guam as our case study. We model pedestrian travel times out of a modeled inundation zone for a local tsunami generated by a M w 8.3 earthquake within the Mariana subduction zone. Evacuation-modeling results indicate that reducing departure delays has a larger impact than increasing travel speeds or creating evacuation corridors through heavy brush on reducing the number of at-risk individuals with insufficient time to evacuate. Travel times to safety are less than wave-arrival times for almost all at-risk individuals in the tsunami-hazard zone if one assumes all three interventions are implemented.

Guam↗

USGS2023

Camera trap distance sampling survey design, Andersen Airforce Base, Guam

Reliable population estimates of animal density is one of the most elementary needs for the control and management of wildlife, particularly for introduced ungulates on oceanic islands. On Guam, Philippine deer (Rusa marianna) and wild pigs (Sus scrofa; wild boar and descendants of domestic pigs) cause agricultural and ecological damage and are hunted for recreational, nutritional, and cultural uses. Most common population estimation methods are based on capture-recapture and related methods that require marking or uniquely identifying individuals. Capturing, marking, and either recapturing or resighting individuals repeatedly is labor intensive and expensive. In many situations marking or individually distinguishing animals is not feasible, necessitating estimating densities and abundance from unmarked animal populations. Motion-triggered camera traps are a relatively low-cost approach that can be used to generate presence/pseudo-absence and indices of relative abundance on multiple species simultaneously. We used distance sampling with camera traps to estimate deer and pig densities from non-independent observations of unmarked animals while accounting for imperfect detection where some present individuals are not detected. We present methods to (1) process the digital imagery data automatically for species detection and species categorization using a machine learning algorithm, (2) automatically estimate distance to detected species using a separate machine learning algorithm, and (3) estimate densities using distance sampling with camera trap methods. We compare accuracy statistics and results of ungulate densities estimated from automated methods to those estimated from manual assessment. We collected 7,695 videos: 381 videos contained deer and 377 contained pigs. The object detection and identification model performed well with overall accuracy above 80% and F1 scores above 0.9. The hazard-rate key detection function was chosen for deer and pigs based on Akaike’s information criterion accounting for overdispersion. Deer density estimates were 0.53 ± 0.20 deer/ha with higher density in the Plateau area than the Tarague area of Guam. Pig density estimates were 0.53 ± 0.32 pigs/ha, also with higher densities in the Plateau area than the Tarague area. Coefficients of variation ranged from 0.38 to 1.15, and greater numbers of camera traps would be required for pigs than deer to achieve desired coefficients of variation. On average, 101.9 ± 82.3 deer and 131.6 ± 118.8 pigs were detected per day. Microsite heterogeneity affected densities where orientation-specific estimates were less precise than estimates made with the full dataset. We developed a camera trap survey design based on standard camera trapping sampling protocols using motion-activated, digital cameras and determined that distance sampling methods using camera traps produce reliable densities of unmarked deer and pigs on Guam. Our camera trap survey design is based on a regularly sized trapping grid that is generalizable and can be expanded to survey other areas of Guam.

Guam↗

USGS2022

Predation thresholds for reintroduction of native avifauna following suppression of invasive brown treesnakes on Guam

The brown treesnake (BTS) ( Boiga irregularis ) invasion on Guåhan (in English, Guam) led to the extirpation of nearly all native forest birds. In recent years, methods have been developed to reduce BTS abundance on a landscape scale. To help assess the prospects for the successful reintroduction of native birds to Guåhan following BTS suppression, we modeled bird population persistence based on their life history characteristics and relative sensitivity to BTS predation. We constructed individual-based models and simulated BTS predation in hypothetical founding populations for each of seven candidate bird species. We represented BTS predation risk in two steps: risk of being encountered and risk of mortality if encountered. We link encounter risk from the bird's perspective to snake contact rates at camera traps with live animal lures, the most direct practical means of estimating BTS predation risk. Our simulations support the well-documented fact that Guåhan's birds cannot persist with an uncontrolled population of BTS but do indicate that bird persistence in Guåhan's forests is possible with suppression short of total eradication. We estimate threshold BTS contact rates would need to be below 0.0002–0.0006 snake contacts per bird per night for these birds to persist on the landscape, which translates to an annual encounter probability of 0.07–0.20. We simulated the effects of snake-proof nest boxes for Sihek ( Todiramphus cinnamominus ) and Såli ( Aplonis opaca ), but the benefits were small relative to the overall variation in contact rate thresholds among species. This variation among focal bird species in sustainable predation levels can be used to prioritize species for reintroduction in a BTS-suppressed landscape, but variation among these species is narrow relative to the required reduction from current BTS levels, which may be four orders of magnitude higher (>0.18). Our modeling indicates that the required predation thresholds may need to be lower than have yet been demonstrated with current BTS management. Our predation threshold metric provides an important management tool to help estimate target BTS suppression levels that can be used to determine when bird reintroduction campaigns might begin and serves as a model for other systems to match predator control with reintroduction efforts.

Ecological Applications↗