Roost Culling as an Effective Management Strategy for Reducing Invasive Rose-Ringed Parakeet (Psittacula Krameri) Populations

C. Jane Anderson (  ecojaneanderson@gmail.com ) Texas A&M University Kingsville https://orcid.org/0000-0002-7472-4242 Leonard Brennan Texas A&M University Kingsville William Bukoski US Department of Agriculture Steven Hess US Department of Agriculture Clayton Hilton Texas A&M University Kingsville Aaron Shiels US Department of Agriculture Shane Siers US Department of Agriculture Bryan Kluever US Department of Agriculture Page Klug US Department of Agriculture


Introduction
Rose-ringed parakeets (Psittacula krameri) are one of most widespread and detrimental invasive avian species (Jackson 2021). The popularity of this species in the pet trade has led to introductions worldwide (Menchetti et al. 2016). Escape from the pet trade, combined with its ability to survive in human-modi ed habitats and tolerance of a wide range of environmental conditions, has led to established populations of rose-ringed parakeets on every continent except Antarctica (Menchetti et al. 2016;ISSC 2021).
Throughout much of their introduced range, populations are increasing; Balmer et al. (2013) estimated the global rose-ringed parakeet breeding range increased over 440-fold from the late 1960s to mid-2010s, making it one of the most rapidly spreading avian species worldwide.
Introduced populations of rose-ringed parakeets have led to negative impacts on native natural resources, economies, and human health and safety. Ecological impacts of invasive rose-ringed parakeets include potential habitat alteration, invasive seed spread, resource competition, and injury and death of native wildlife from the parakeet's highly antagonistic behavior (Shiels and Kalodimos 2019). In Australia, invasive rose-ringed parakeets strip tree bark which has led to tree death (Fletcher and Askew 2007). Shiels et al. (2018) determined 97% of invasive rose-ringed parakeets evaluated in Hawai'i, USA, had consumed yellow guava (Psidium guajava), potentially leading to seed dispersal of this invasive plant. In Europe, native birds responded to rose-ringed parakeet presence by reducing feeding rates and increasing vigilance (Peck et al. 2014). Perhaps the greatest impact on native species stem from rose-ringed parakeets' nesting behavior. Rose-ringed parakeets are secondary cavity nesters, utilizing cavities that are naturally occurring or created by other species. In Israel, cavity enlargement by rose-ringed parakeets appears to facilitate breeding success of invasive common mynas (Acridotheres tristis; Orchan et al. 2012). In Spain, invasive rose-ringed parakeets maim and kill imperiled greater noctules (Nyctalus lasiopterus) to claim cavities, which has led to decreased abundance of the bats (Hernandez-Brito et al. 2018).
Beyond ecological impacts, invasive populations of rose-ringed parakeets pose human health and safety threats. They are noted agricultural pests in their native and introduced ranges, as they frequently depredate fruit, grain, and seed crops (Khan 1983, Khan et al. 2011, Klug et al. 2019). Rose-ringed parakeets congregate nightly in roosts, which are often in urban, peri-urban, or agricultural areas.
Excessive droppings at roost sites potentially expose humans to zoonotic pathogens such as psittacosis and avian u (Klug et al. 2019). Furthermore, aircraft have struck rose-ringed parakeets at London's Heathrow Airport on three documented occasions, which posed a threat to human safety and cost >£20,000 British pounds each (Fletcher and Askew 2007).
Given the rapidly increasing distribution of rose-ringed parakeets and the pervasive negative impacts of this species, natural resource managers throughout the world are seeking effective population reduction strategies. In parts of their native range, rose-ringed parakeet populations are thought to be reduced by capture for the pet trade (Menchetti 2016). To  Rose-ringed parakeets were introduced to the island of Kaua'i, Hawai'i, USA, in the 1960s. The population remained relatively low for the following four decades, demonstrating a lag period that is common  Beginning in 2020, a new management effort was implemented to control the rose-ringed parakeet population on Kaua'i by culling birds at their roosts; this effort was conducted by an independent wildlife control company and funded by the County of Kaua'i. Prior to this effort, roost culling had been generally avoided on Kaua'i for ecological and social purposes. Primarily, there was concern culling would lead to roost abandonment, which would be problematic because roosts are the primary method of monitoring the abundance of this species. Further, because the roosts were in developed areas, managers were concerned roost abandonment would lead to new roosts in natural areas; this could increase impacts on native wildlife species, which are largely restricted to central portions of the island. Second, two of the three known roosts were adjacent to tourist resorts, and there was concern of how the public would perceive culling efforts. In this regard, reduced tourist activity due to the COVID-19 pandemic provided a unique and optimal timeframe to trial roost culling as a rose-ringed parakeet population management strategy. Only ~331,000 tourists traveled to Kaua'i in 2020, compared to 1.37 million in 2019 (Hawai'i Tourism Authority 2021). Further, Kaua'i implemented a 21:00-05:00 curfew from March 20 -May 6, 2020, as a COVID-19 safety precaution (County of Kaua'i 2020). The reduced human activity allowed nighttime roost culling to be conducted in public areas with relatively few public observers.
We evaluated overall impact of roost culling on estimated island-wide minimum rose-ringed parakeet abundance. We further evaluated take per hour of shooter effort, estimated the age and sex ratio of animals removed, and monitored whether the parakeets abandoned their roosts in response to culling efforts. Findings can be used to inform implementation and best practices of roost culling for management of nonnative rose-ringed parakeet populations.

Study Species
Rose-ringed parakeets are considered a medium-size parrot, with adult body mass ranging from 131-180

Study Area
The Hawaiian island archipelago is among the most remote in the world. It consists of eight major islands and dozens of smaller islets. Kaua'i is the northernmost and fourth largest of the major islands ( Figure 1). It has an area of 1,430 km 2 and ranges in elevation from sea level to 1,598 m. Kaua'i has a tropical climate, with average daily temperatures of 26° C in February and 29° C in August. January-March is regarded as the wet season, although precipitation is high year-round; mean annual rainfall ranged from 440 mm in the lowlands to nearly 10,000 mm at the highest peak (Giambelluca et al. 2013).
Agriculture impacted by rose-ringed parakeets on Kaua'i includes large hybrid seed corn farms on the southwestern portion of the island (converted from previous sugar cane elds) and tropical produce farms throughout the lowlands including lychee, mango, papaya, avocados, longan, and citrus, among others.

Study Design
We estimated the island-wide minimum abundance of rose-ringed parakeets on Kaua'i prior to, during, and post roost culling through roost counts, the method most commonly used to estimate rose-ringed parakeet population size (Butler 2003, Pithon and Dytham 1999, Shiels and Kalodimos 2019; Figure 1). Roost counts were conducted by counting rose-ringed parakeets as they departed the roost at sunrise.
When ocks were too large to count by individual, parakeets were counted by increments of ve or ten individuals (Shiels and Kalodimos 2019). Each roost count was conducted two to four times within a twoweek period, and counts were averaged to represent the respective estimate (Table 1; Figure 1). This technique should be considered a conservative population estimate, or minimum number of known individuals, as it is possible there were unknown roosts during any given time; thus, we refer to cumulative estimates from roost counts as island-wide minimum abundance.
Prior to roost culling we estimated island-wide minimum abundance in January 2020, at which time there were three known roosts: Po ipū, Lihue_1, and Kapa'a ( Figure 1). There was a pause in roost culling from November 2020 to February 2021; we estimated island-wide minimum abundance twice during this period, once immediately after culling cessation in November 2020 and once immediately before culling was resumed in February 2021. In November 2020 there were four known roosts: Po ipū, Lihue_2, Koloa, and Kapa'a. In February 2021 there were four known roosts: Po ipū, Koloa, Lihue_3, and Kapa'a. In April 2021, we estimated minimum island-wide abundance after the completion of culling at which time there were four known roosts: Po ipū, Lihue_2, Kapa'a, and Princeville ( Figure 1). represent estimated reproductive season. Reported roost counts represent estimated island-wide minimum abundance (Table 1).
Thirty nights of roost culling were conducted from March 2020 -March 2021; these were divided over three phases ( Figure 1). Phases were differentiated by whether they included the estimated nesting season, number of roosts culled, and culling frequency (see below). For each night of culling, we documented the estimated number of parakeets lethally removed, the total number of carcasses retrieved, the number of shooters, and the number of hours culling was conducted. Not all carcasses were retrieved, as some birds ew away with injuries and some carcasses were inaccessible when caught in vegetation or scavenged by feral cats (J. Young, Kani Wildlife Control, personal communication); thus, the discrepancy in removal estimates and total carcass retrieval. All data -estimates of parakeets removed nightly, number of carcasses retrieved, number of shooters, and number of hours worked -were provided by the private company that implemented the culling effort, Kani Wildlife Control (Kekaha, HI, USA).
Culling was conducted by one or two shooters each night using sound-suppressed 22-caliber air ri es with telescopic sights. Each shooter had a partner using a white spotlight to locate parakeets perched in palm trees. We calculated the total shooter hours for each night as the number of hours worked per shooter (i.e., two shooters working for three hours = six shooter hours). We calculated the estimated number of birds culled and total number of carcasses retrieved per shooter hour. We calculated the means (± standard deviations) of these values for each phase.
Roosts subjected to culling: The roosts of rose-ringed parakeets on Kaua i were in urban to peri-urban habitats. We measured the approximate size of each roost by documenting the coordinates of the two furthest points of the roost and measuring the distance between them using ArcMap (version 10.6.1, ESRI, Redlands, CA, USA). The Po ipū roost included >100 ornamental trees over approximately 1 km; trees used for roosting by the roseringed parakeets included royal palms (Roystonea regia) and king palms (Archontophoenis alexandrea). Within this area were four tourist resorts, the edge of a golf course, a public park, and >50 private residential properties. The Lihue_1 roost included 46 royal palms and two coconut palms (Cocos nucifera) concentrated in the parking lot of a shopping center; total length of the roost was ~190 m. The Lihue_3 roost was not occupied at the beginning of the study; it had historically been a roost site less than six months prior to the study (CJ Anderson, personal observation) and was repopulated by the parakeets between November 2020 -February 2021. This roost was spatially the smallest of all evaluated, with a total length around 60 m. In this site, the parakeets were using ~20 royal palms in a lawn surrounding a convention center. The Koloa roost was not occupied at the beginning of the study but was also historically a roost site, with documented use in 2011 (Gaudioso et al. 2012). The parakeets re-established the Koloa roost around June 2020 (WP Bukoski, personal observation) in a row of ~50 palm trees along a road between a residential street and a golf course. The length of the roost was approximately 237 m ( Figure 1).

Phase 1: March -May 2020
Twenty-one nights of culling were conducted from March 27 -May 29, 2020, between two roosts, Lihue_1 and Po ipū (Figure 1). No culling was conducted at the third known roost at the time, the Kapa'a roost. The Lihue_1 roost was markedly smaller in estimated number of birds occupying the roost and spatial length than the Po ipū roost (see above implemented at an average rate of every 2.5 days (min = 1, max = 6).
We obtained 2,993 carcasses from the parakeets culled in Phase 1. We documented the total number of adult males, as noted by their black and rose-colored neck collars. Birds lacking neck rings were initially classi ed as unknown age and sex. From the sample of unknown birds, we conducted 601 necropsies to determine age and sex in three classes: sexually mature females, immature females, and immature males. We determined sex by presence of testes or ovaries. Immature males were classi ed as those with testes but lacking neck rings. We determined female age by follicular development; females in which all follicles were undeveloped were designated as sexually immature, and those with active follicles were designated as sexually mature (J. Heatley, Texas A&M University, personal communication; Figure S1 in Supplementary Information). Of the 601, 16 carcasses had too much internal damage to accurately assess age and sex, leading to a usable sample of 585 carcasses (32% from the Lihue roost; 68% from the Po ipū roost). We assumed the age and sex ratio of the unknown birds we did not necropsy (n = 1,755) was the same as those we necropsied. We therefore used the proportion of each age and sex class among necropsied birds to estimate the proportion of the respective age and sex class in the total sample (n = 2,993). Using these ndings, we estimated the proportion and 95% con dence interval of each age and sex class of parakeets removed during the entire study period.
Phase 2: June -November 2020 Phase 2 was a continuation of culling efforts at the Po ipū roost ( Figure 1). We evaluated this time period separately as it did not include the predicted nesting season and differed in culling frequency. There were six nights of culling between June 5 November 12, 2020 (27 shooter hours). Frequency of culling effort was less consistent during this time as compared to Phase 1, with an average of 32 nights between culling efforts (min = 1, max = 85). The largest lag between culling events was from August to November 2020 (85 days).

Phase 3: February & March 2021
Two nights of culling were implemented at the Lihue_3 roost on consecutive nights, February 23 and 24, 2021 (17.5 shooter hours; Figure 1). To evaluate the impact of successive disturbance on roost abandonment, we conducted roosts counts at the Lihue_3 roost on February 22, 24, and 26, 2021. For all carcasses collected at the Lihue_3 roost, we recorded the number of adult males and number of unknown age and sex birds. From the rst of the two nights of culling, we necropsied 52 of the unknown birds to estimate the age and sex ratio of the parakeets removed. We assumed the age and sex ratio of carcasses we did not necropsy (n = 165) were the same as those we did.
The nal night of roost culling was conducted at the Koloa roost ( Figure 1) in March 2021 (9.32 shooter hours). The company conducting the culling reported the estimated number of parakeets removed (n = 214) and reported total number of carcasses retrieved (n = 165) but did not obtain carcasses to evaluate age and sex.

Comparison of age/sex culled by season
We compared age and sex ratio of rose-ringed parakeets culled by season. During the nesting season, nesting female rose-ringed parakeets remain on their nests rather than returning to the roost at nightbeginning when the rst egg is laid (Braun and Wink 2013) -and are therefore not expected to be available for roost culling. Nesting season of rose-ringed parakeets on Kaua'i has not been documented. However, nesting season of rose-ringed parakeets on the neighboring island O'ahu has been observed as late January -mid April (Shiels and Kalodimos 2019). We therefore evaluated whether age and sex ratio differed between birds removed in March and April 2020 (n = 1,140) with those in May 2020 (n = 1,853) during Phase 1. While February is within the predicted nesting season, post-hoc observation indicated a higher proportion of adult females was culled in February 2021 (n = 402). We performed a multinomial logistic regression using the nnet package in R (version 4.0.5) with age/sex as the response variable (mature female, mature male, immature female, and immature male) and culling season as the independent variable (March/April 2020, May 2020, February 2021) to evaluate whether age and sex ratio of parakeets culled varied signi cantly by season (code provided in Supplementary Information).

Results
We estimated an island-wide minimum abundance of 10,512 rose-ringed parakeets in January 2020 (prior to roost culling) among the three known roosts (Lihue_1, Po ipū, and Kapa'a; Table 1).   Over 21 nights of culling, an estimated 4,926 rose-ringed parakeets were removed from the Lihue_1 and Po ipū roosts, a con rmed 3,619 carcasses were retrieved, and average take per shooter hour was 55.0 (SD = 23.5). At the Lihue_1 roost, an estimated 561 parakeets were removed (530 carcasses retrieved) over four nights of culling, approximately 22% of the pre-culling roost abundance (n = 2,559; Table 1). During the fourth night of culling, the number of birds removed per shooter hour decreased substantially (13.1) from the average of the rst three nights (59.8), as the parakeets had partially abandoned the roost (J. Young, Kani Wildlife Control, personal communication). After the fourth night of culling, the parakeets fully abandoned the roost.
An estimated 4,365 parakeets were removed from the Po ipū roost -approximately 68% of the pre-culling roost abundance (6,452; Table 1) -and 3,089 carcasses were retrieved. Average take per shooter hour was 57.0 (SD = 18.6). Over the 17 nights of culling, the parakeets did not fully abandon the roost. While we did not measure the number of birds per tree or per individual property, we noted that the extent and composition of the roost changed. For example, during this period, the parakeets shifted approximately 175 m west onto a different property. In many areas, the parakeets shifted to private properties where culling was not permitted (J. Young, Kani Wildlife Control, personal communication).
Of the 2,993 carcasses evaluated, 653 (21.8%) were sexually mature males. Of the 585 carcasses necropsied, 142 were immature males, 123 were mature females, and 320 were immature females (Table   3). Using these ndings to estimate the total samples size, we estimated the 4,926 parakeets removed during Phase 1 included 21.8 ± 3.4% mature males, 19.0 ± 3.2% immature males, 16.4 ± 3.0% mature females, and 42.7 ± 4.0% immature females ( Figure 2).  Table 2) was lower than Phase 1 and ranged from 7.8 to 13.7 in June-August. After an 85-day hiatus in culling, the number of birds removed per shooter hour increased in November to 82.3 ( Figure 3). We were unable to evaluate age and sex of carcasses retrieved during this time due to COVID-19 travel restrictions.
We estimated the island-wide minimum abundance was 4,963 individuals among four roosts in November 2020 after the conclusion of Phase 2 roost culling. Given the Lihue_1 roost was abandoned during roost culling, there were no birds in this location. The Po ipū roost was greatly reduced in size due to culling and partial roost abandonment. The Kapa'a roost, where no culling occurred, had increased by approximately 27%. Two roosts had established, Lihue_2 and Koloa; it was unknown when the Lihue_2 roost established, but the Koloa roost was rst documented in June 2020 ( Figure 1; Table 1; W. Bukoski, personal observation). While neither roost was present in January 2020, both were historic roost locations that had previously been abandoned for unknown reasons. Between November 2020 and February 2021, the Lihue_2 roost abandoned for unknown reasons and the Lihue_3 roost became established (Figure 1). Among the four known roosts in February 2021, we estimated island-wide minimum abundance to include 6,883 rose-ringed parakeets (Po ipū, Lihue_3, Kapa'a, and Koloa; Of the 402 carcasses retrieved, 185 (46.0%) were mature males. Of the 52 birds of unknown age and sex necropsied, we identi ed 28 mature females, nine immature females, and 15 immature males (Table 3). Using these observations, we estimated take from the Lihue_3 roost during this period included 29.1 ± 3.7% sexually mature females, 15.6 ± 2.9% immature males, and 9.3 ± 2.4% immature females ( Figure 2). Qualitatively, we observed that follicles were more developed among females in February 2021 than March-May 2020, including one female with an intact egg.
During the single night of culling at the Koloa roost, an estimated 214 individuals were removed (165 retrieved carcasses). The parakeets did not return the following evening, indicating they abandoned the site after only one night of culling. An estimated 18.6% of the roost was removed prior to abandonment. Table 3 Observed age and sex of rose-ringed parakeets (Psittacula krameri) removed via roost culling on Kaua'i, Hawai'i, USA, in March -May 2020 (n = 2,993, Phase 1) and February 2021 (n = 402, Phase 3). We identi ed all mature males by plumage, whereas we identi ed a sample (n = 585, Phase 1; n = 52, Phase 3) of mature females, immature females, and immature males via necropsy. Carcasses were not evaluated during Phase 2 due to COVID-19 travel restrictions. The age and sex ratio of rose-ringed parakeets culled varied between the March-April 2020 (n = 1,140), May 2020 (n = 1,853), and February 2021 (n = 498) sampling periods (Table 3; Figure 2). The likelihood of harvesting a mature female vs an immature female, immature male, or mature male was signi cantly higher in February compared with March/April or May (p <0.01; Table 4). Discussion Roost culling appears to be an e cient mechanism to rapidly remove large numbers of rose-ringed parakeets with relatively little effort. In this study, an estimated >6,000 parakeets were removed in one year at a cost of around $30,000 US (J. Young, Kani Wildlife Control, personal communication). While the required effort and cost of this method were relatively low, roost culling appears to lead to full or partial roost abandonment, which can have negative consequences including compromising population monitoring (Bunbury et al. 2019) or potentially disturbing native species.
All roosts subjected to roost culling in this study were either fully or partially abandoned by rose-ringed parakeets. The likelihood and rate of abandonment appeared to be related to roost size as well as exposure of parakeets to previous roost culling. Two roosts were culled in Phase 1 and therefore included parakeets without prior exposure to roost culling. The Lihue_1 roost was partially abandoned after three nights of weekly roost culling and fully abandoned after four nights. The Po ipū roost was never fully abandoned; however, this roost was spatially restructured and partially abandoned by the parakeets. The lack of complete abandonment may have been due to the large size of the Po ipū roost, which allowed shooters to operate in sections of the roosts (e.g., individual properties) rather than disturbing the entire roost simultaneously (J. Young, Kani Wildlife Control, personal communication). The two roosts culled in Phase 3 (Koloa and Lihue_3) were both established after other roosts had been abandoned in response to roost culling. Thus, it is feasible the birds in these roosts came from the abandoned roosts and therefore had previous exposure to roost culling. Parakeets in these roosts fully abandoned after one and two nights, respectively. Collectively, these ndings indicate roost culling is best used for roosts spatially large enough (~ 1 km) to cull in sections, or small enough in abundance to cull entirely in no more than two nights. The optimal frequency of roost culling implementation, including whether consecutive nights of culling should be utilized, is di cult to ascertain from this study. Future research should further investigate the impact on culling frequency on roost abandonment.
We found an average of ~45 rose-ringed parakeets can be culled per shooter hour via roost culling; managers can use this estimate to determine whether roosts are small enough and staff size is su cient to successfully remove all birds from the roost prior to abandonment. While the Po ipū roost was not fully abandoned despite culling throughout Phases 1 and 2, take per shooter hour substantially decreased during Phase 2 ( Figure 3); this nding indicated that parakeet exposure to roost culling or decreased roost size may in uence e cacy of this strategy. It is important to note that the shooters in this study were professional wildlife control agents with many years of experience; less experienced shooters would likely have a lower take rate.
The island-wide minimum abundance of rose-ringed parakeets on Kaua'i decreased by an estimated >3,000 birds from between January 2020 -April 2021 (Table 1) indicates the 2020 abundance estimate in this study was likely conservative. Continued research to locate rose-ringed parakeet roosts and more precisely estimate population size is merited.
The number, locations, and sizes of roosts varied during the study. Because roost counts are the primary method used to estimate parakeet abundance on the island, roost inconsistency complicates population monitoring. Rose-ringed parakeets will sometimes abandon roosts for unknown reasons in the absence of management (Peck 2013, Pyle andPyle 2017). Roost abandonment and restructuring appeared to occur during this study both as a direct consequence of roost culling (i.e., Lihue_1, Lihue_3, and Koloa roosts) as well as for unknown reasons (i.e., Lihue_2; Figure 1) Understanding the biology and population dynamics of a target species is critical for population management. An effective management plan for invasive species population reduction requires an informed strategy including information on target numbers to be culled as well as the spatial distribution, duration, and timing of lethal removal (Grarock et al. 2014). It is unclear why the proportion of juvenile females removed in this study was higher than juvenile males. Little information exists on the in uence of social structure or hierarchy on physical placement of rose-ringed parakeets within or among roosts; it is possible juvenile females are located in areas of the roost that are easier to target than other age or sex classes or that division among roosts is related to age and sex. It is also possible juvenile females represent a larger proportion in the population, perhaps due to higher survivorship, although to our knowledge this dynamic has also not been explored. With long-lived bird species, removing sexually mature individuals can have a greater impact on population suppression than removal of juveniles (Ellis and Elphick 2007). While many parrot species demonstrate monogamy (Spoon 2006), rose-ringed parakeets do not form life-long pair bonds; this may suggest culling adult females has a greater impact on survivorship and fecundity than culling adult males. In this study, we found a larger proportion of sexually mature females among birds culled at the roost in May 2020 and February 2021 as compared to March and April 2020; this may suggest future rose-ringed parakeet roost culling efforts on Kaua'i should be restricted to months outside of March and April. Further research is merited to identify the breeding season of rose-ringed parakeets on Kaua'i as well as at-nest metrics, such as clutch size and nest success rates. Population modeling could further evaluate the e cacy of culling each age and sex class and e cacy of roost culling versus management at the nest.
As the rose-ringed parakeet distribution spreads and populations increase, managers will continually need informed management strategies to control populations. While eradication has proven successful for small, incipient populations (Bunbury et al. 2019; Saavedra and Medina 2020), it may not be feasible for most invasive rose-ringed parakeet populations. Roost culling appears to be an e cient, albeit imperfect management strategy to reduce rose-ringed parakeet populations and perhaps suppress populations below target thresholds. Like other management tools, effective implementation of roost culling will need to be sustained, strategic, and implemented long-term. Culling should also be coupled with evaluations of population growth rate to determine e cacy. Future research of roost culling should more speci cally evaluate the in uence of culling frequency (e.g., culling on consecutive vs. nonconsecutive nights at similar-size roosts) as well as using alternative options to white spotlighting (e.g., infrared). While further research is merited, it appears roost culling is a viable option to incorporate into integrated pest management programs for this invasive species.   (Table 1).

Figure 2
Estimated number and percent (95% CI) of rose-ringed parakeets (Psittacula krameri) by age and sex removed via roost culling from March -April 2020 (n = 1,511 from Lihue_1 and Po ipū roosts), May 2020 (n = 3,415 from Po ipū roost), and February 2021 (n = 498 from Lihue_3 roost) on Kaua'i, Hawai'i, in 2020 and 2021. We identi ed all mature males by plumage, whereas we identi ed a sample of mature females, immature females, and immature males via necropsy during Phase 1 (n = 337 from March -April; n = 248 from May) and Phase 3 (n = 52). Carcasses were not evaluated during Phase 2 due to COVID-19 travel restrictions.