From colony to fallout: Artificial lights pose risk to seabird fledglings far from their natal colonies

Seabirds are at risk of decline from multiple threats, including artificial light resulting in their grounding (“fallout”). Without evidence, it can be assumed that seabirds travel short distances from natal colonies to their fallout location, potentially to the closest light source. To test this, a case study on wedge‐tailed shearwaters (Ardenna pacifica, ʻUaʻu kani) fledging from the island of Oʻahu, Hawaiʻi was conducted. To assess fallout locations in relation to natal colonies, we affixed identification bands to 4648 chicks at nine natal colonies prior to fledging from 2018 to 2022. Distances between fallout location and natal colony were mapped and fallout location characteristics (radiance, elevation, distances to nearest coastline, road, and colony) were analyzed for 27 banded fledglings that were discovered post‐fallout. The distance between the natal colony and fallout location was significantly greater than the distance between the nearest colony and the fallout location. Fallout often occurred on opposite coastlines from the natal colonies and at substantial distances (x¯ = 24.91 km), with one fallout event recorded on a different island from the natal colony. Our results demonstrate that all artificial lights, regardless of proximity to seabird colonies, may pose a threat to seabird fallout and that fallout is often occurring far from natal colonies and in inconsistent patterns. Collaborative, large‐scale, and multi‐island light management alteration, particularly during fledging periods, is critical in recovering Pacific Island seabird populations. Actions such as these will benefit not only light‐sensitive species such as seabirds and sea turtles but also human health and night‐sky viewing.


| INTRODUCTION
Seabirds are one of the most threatened and rapidly declining groups of vertebrates (Croxall et al., 2012;Dias et al., 2019), with declines of 70% between 1950 and 2010 (Paleczny et al., 2015).Alongside predation by invasive species in breeding areas, declining breeding and nesting habitats, diminished food resources, and impacts from plastics and fishing gear, artificial lights greatly contribute to declines in seabird populations (Harrison, 1990).This "light pollution" has affected many ecological processes including migration, foraging, chick provisioning, survival, and reproduction (Cianchetti-Benedetti et al., 2018;Gaston et al., 2014;Klomp & Furness, 1992;Raine et al., 2017;Troy et al., 2011).This problem has grown over time following the increase in the magnitude of artificial lighting due to industrialization and human population growth over the last century (Gaston et al., 2012).
The threat of artificial lights to seabirds appears to be greatest during fledging, the act of taking flight from the natal colony for the first time (Atchoi et al., 2020;Rodríguez, Holmes, et al., 2017).During this first flight from the nesting colony toward ocean feeding grounds, artificial lights may attract, disorient, or confuse fledging seabirds, often due to an innate attraction to light (Montevecchi, 2006).Whether by exhaustion, confusion, or collision with structures, this can result in the grounding of seabirds, a phenomenon termed "fallout" (Montesdeoca et al., 2017;Rodríguez, Moffett, et al., 2017).Once grounded they are vulnerable to threats of automobile collisions, physical injury, predation, dehydration, hypothermia, and starvation, and are often unable to take flight again depending on location and weather (Ainley et al., 2001;Baccetti et al., 2005;Deppe et al., 2017;Rodríguez et al., 2014;Smith et al., 2002).
Due to the diverse factors that can contribute to fallout, conservation outcomes may be improved by a better understanding of the relationships between fallout location and the natal colony.Artificial lights increase seabird mortality in fledglings during their initial flights from their natal colony (Croxall et al., 2012;Fontaine et al., 2011;among others), but questions remain regarding the distance at which sources of light may impact fledging seabirds and how fallout is oriented spatially in relation to the natal colony.Previous studies have explored spatial patterns of fallout, but many have lacked either the identification data necessary to associate fallout birds to their colonies of origin, or sample sizes to conclude population-level patterns (Deppe et al., 2017;Friswold et al., 2020;Troy et al., 2013).Troy et al. (2013) used GIS-based modeling from fallout records in Hawai'i to infer light attraction patterns and Rodríguez et al. (2015) were the first to address spatial distribution preceding and following fallout-particularly the relationship between natal colony location and fallout sites.Some studies have used GPS transmitters to track seabird movements (Rodríguez et al., 2015;Wilson et al., 2002; among others), but the number of individuals in such studies limited implications for broad fallout patterns and did not always directly assess the impact of light pollution.
The application of federally-distributed metal identification bands on seabirds offers an affordable approach to evaluating dispersal and fallout patterns at a larger scale.Banding is the process where plastic or metal bands with a unique reference number and/or color are affixed to a bird's leg(s) (Weimerskirch et al., 1985).Banding seabird chicks before fledging makes it possible to identify potentially cryptic spatial patterns that occur following fledging for individuals that experience fallout, especially when using an abundant model species.Assumptions can be made regarding the orientation between fallout location and natal colony, but unless individuals are marked at the natal colony prior to fledging, this assumption cannot be tested.Previous studies have affixed bands and/or transmitters on recovered birds following fallout events but not before the fledging period, in which case the colony of origin cannot be confirmed (le Corre et al., 2003;Raine et al., 2020).
To address research gaps regarding the impact of artificial lights on seabirds in the Hawaiian Islands, a model species was selected that was relatively abundant in the study area, allowing for a large-scale banding study.Wedge-tailed shearwaters (Ardenna pacifica, WTSH, ʻUaʻu Kani), in the order Procellariiformes, are one of the most abundant seabird species in the Hawaiian Islands (Pyle & Pyle, 2017), and as such, are the largest contributor to seabird fallout across the islands (Work & Rameyer, 1999).Thus, results from studies of WTSH are potentially informative for conservation of more threatened species.WTSH are most abundant at sites free of invasive predators, typically offshore islets and/or within mammal exclusion fences and areas (Pyle & Pyle, 2017;USFWS, 2005).The first predatorproof fence in the Hawaiian Islands is at Kaʻena Point Natural Area Reserve on the island of Oʻahu.Since its construction, the WTSH population has increased substantially at the site (VanderWerf et al., 2014).Nonprotected colonies occur in beach parks and humandominated areas with a diversity of predators and impacts from human trampling (Idle et al., 2021).
WTSH occupy an important role in coastal ecosystems, with their guano acting as a nutrient source for coral reefs and native coastal vegetation (Honig & Mahoney, 2016).Like all seabirds, they are ecosystem indicators, high-order predators, and beneficial to marine ecosystem health (Smith et al., 2011).After hatching in the summer months, WTSH fledge from early November through late December (Whittow, 1997) with peak fallout on Oʻahu occurring in mid-November (Friswold et al., 2020).WTSH nest in regions of varying exposure and proximity to anthropogenic threats, including artificial light, making them an exemplary model species for studying fallout patterns (Grady, 2020;Pettit et al., 1984;USFWS 2005).
In this study, we tested a common assumption-that most fallout individuals originate from colonies close to their fallout location (Friswold et al., 2020;Rodríguez et al., 2015).In addition to testing this hypothesis, we sought to understand the degree of variability in distances of natal colonies to fallout locations, and other fallout location characteristics such as elevation, radiance, and distances to nearest WTSH colony, coastline, and road to inform management interventions and future research.

| METHODS
In this study, a robust and extensive banding effort (necessary given the low rate of recapture for fallout birds) of WTSH chicks occurred during the nesting phase prior to fledging.The research was undertaken from 2018 to 2022 on the island of O'ahu at nine WTSH colonies.
After fledging, the locations of banded individuals that were grounded and recovered were analyzed.Linear flight paths and distances of fallout locations in comparison to the natal colony were assessed.The fallout location characteristics that were assessed for the recovered banded WTSH fledglings included distance to the natal colony, distance to the nearest colony, distance to the coastline, distance to the nearest road, radiance value of the fallout location, and elevation of the fallout location.Our analyses did not include information on un-banded fallout individuals, nor the totality of fallout on the island, since it was not possible to determine the colony of origin for un-banded birds, and the main intention of this research was to assess the relationship between fallout location and natal colony.

| Study area
The study was conducted on the island of Oʻahu, Hawaiʻi, the most populated and industrialized of the Hawaiian Islands, and thus the island with the highest light radiance values (Green et al., 2022).Radiance is the light emitted or radiated off of a surface; high radiance values indicate heavily light-polluted areas (Cinzano & Falchi, 2014;Troy et al., 2011).The island of Oʻahu contains 18 known WTSH breeding colonies and nine suspected colonies with approximately 80,000 breeding pairs total (Figure 1; Pyle & Pyle, 2017;Urmston, 2020).Colonies exist mostly on mammal-eradicated offshore islets and along the shoreline of many Hawaiian Islands (Smith et al., 2006;USFWS, 2005).
Banding occurred at nine WTSH colonies that included the visitor-restricted offshore islets of M anana, Mokulua Iki, and K aohikaipu, and the visitor-allowed offshore islets of Popoia and Mokulua Nui, all of which have undergone mammalian predator eradication.The Oʻahu onshore colonies included the un-protected colonies of Kailua Beach Park, Kahuku Golf Course, and the mammal-controlled onshore colonies of Marine Corps Base Hawaii-Kaneohe Bay (MCBH-KB) and Kaʻena Point Natural Area Reserve.Heavily light-polluted areas are typically located around the center and southwest of the island.The colonies selected for banding were predominantly located in the southeastern region of Oʻahu-as this is where the majority of colonies occur on the island (Figure 1).The colonies selected for this study were additionally chosen based on state and governmental allowance as well as the ability to access colonies safely and effectively without a high risk of collapsing burrows while moving through the nesting colony.WTSH chicks were banded approximately two weeks to one month prior to peak fledging in late October and early November (peak fledging occurs in mid to late November), once chicks attained sufficient leg size where bands could not slip over the joint.Between 4-7 hours were spent conducting chick extractions at each site each year.Not all colonies were visited each year due to weather, ocean conditions, and restrictions related to the -COVID-19 pandemic.Chicks were extracted using standard practices: reaching into a burrow, allowing the chick to bite the hand, and then gently pulling the chick out of the burrow by its bill.Chicks were held and banded in a way to minimize stress and the chance of injury.US Geological Survey stainless steel bands used for this study were distributed under the Bird Banding Laboratory.Bands were affixed to the right leg and analyzed for light penetration at the band joint to ensure complete closure.For every bird handled, the presence of avian pox, ants, deformities, and Hippoboscid flies (Hippoboscidae) was recorded and general health status was evaluated to determine if they were of adequate health for banding.If chicks exhibited extreme stress (gular fluttering, overheating, regurgitation, etc.), deformity, underdevelopment, or any of the aforementioned symptoms to a degree that could reduce fitness or survival, they were not banded.

| Fallout recovery data of banded WTSH
We obtained and collated data from banded birds that were found following fallout recovery after each subsequent fledging season.Data were obtained from multiple wildlife rehabilitation centers and fallout response organizations (Hawai'i Wildlife Center, DLNR, DOFAW, Feather and Fur, Sea Life Park, HMAR, MNSRP).For banded WTSH fledglings recovered, the band number, recovery date, and location of recovery (often as an address or street intersection) were collected from the contributing organizations.Fallout recovery came predominantly from general public reports to the bird recovery organizations previously mentioned.One systematic bird search is conducted yearly on the Southeast region of the island by Hawai'i Pacific University, however, none of the banded fallout birds were recovered on these surveys.Because fallout individuals were discovered and collected by varying persons and organizations, it was not possible to know the weather conditions for each area and/or exact dates that fallout occurred and thus this was excluded from the analysis.However, average monthly wind speeds for the month of highest fallout (November) are estimated to be between 0 and 1 knots blowing northeasterly, with peak wind speed in the Hawaiian Islands occurring in June and July with wind speeds between 1 and 7 knots on average (NOAA US Wind Climatology report, 2022).

| Spatial analyses
For each grounded WTSH recovered with a band, the natal colony (nesting colony where banding and subsequent fledging occurred) was determined based on the band number and then the distance between the fallout location and natal colony, the fallout location and the nearest WTSH colony (not necessarily the one they fledged from), the fallout location to coastline, and the fallout location to a major road were calculated.Additionally, fallout location radiance (nW/sr*cm 2 ; 2020 National Oceanic and Atmospheric Association [NOAA] Visible Infrared Imaging Radiometer Suite [VIIRS]), fallout location elevation, and outcome of individuals (euthanized, released, died, or unknown) were also assessed.
Distances between fallout locations and colonies, roads, and coastlines were calculated in ArcGIS (ArcGIS online, 2022) in a data frame with a Web Mercator coordinate system.The measurement tool uses a planar measurement with 2D Cartesian mathematics to calculate length.Distances were calculated as a mean for each fallout recovery event using the greatest and lowest distance between the fallout location and colony.A mean length was used as colonies could span up to 1 km and precise fallout location coordinates were occasionally unknown (often recorded as an address, street name, intersection, or road mile).Thus, distance calculations taken from the center points of fallout locations and colonies varied up to 0.90 km (maximum length of largest colony + maximum length of fallout location).Fallout recoveries lacking defined locations (more than 2 km variability) were omitted.Elevation was calculated using ArcGIS World Atlas 2015 data from either the exact location of fallout recovery or for those in which exact locations were unknown from the center point of the fallout location area (not exceeding 2 km in variability).
Radiance was acquired using 2022 NOAA VIIRS data accessed using the website (lightpollutionmap.info)and the Earth at Night 2016 (Esri, 2014) living atlas layer.This layer presents a nighttime view of the Earth and locations of permanent lights on the Earth's surface as fledging often occurs in the nighttime.The light pollution overlays were created from the monthly "VIIRS DNB Cloud Free Composites" dataset and obtained via the Earth Observation Group and NOAA National Geophysical Data Center.

| Statistical analyses
Statistical analyses were performed using R version 1.1.463(R Core Team, 2018).The results were presented as mean ± SD and range for fallout location characteristics (location radiance; location elevation; distance from fallout location to natal colony; distance from fallout location to the nearest colony; distance from fallout location to coastline; and distance from fallout location to a major road).To test the hypothesis that WTSH fallout in close proximity to their natal colonies, mean values for fallout location to the natal colony and fallout location to the nearest colony were tested for normality using a Shapiro-Wilk test, then log-transformed and compared using a paired t-test.

| Banding and recovery
In total, 4648 WTSH chicks were banded from 2018 to 2022 with zero known WTSH injuries or fatalities as a result of banding activities.A total of 29 banded birds were recovered post-fallout.Two recoveries were omitted due to incomplete fallout location data, resulting in n = 27 for spatial analyses (N = 4648, n = 29, percentage of banded individuals recovered = 0.62%; n = 27, percentage of banded individuals recovered and analyzed = 0.58%).The majority of bands were recovered due to reports of downed birds to wildlife organizations or fallout birds that were discovered by state authorities or local citizens, typically in a non-standardized manner.Recovered bands were discovered in a variety of locations that included the airport, military hangars, along highways, near businesses, urban streets, beach parks, and a research pier (Figures 2,  3, and Table 1).Of the 29 recovered fallout individuals, 23 were released following recovery, four were euthanized due to injury related to fallout, and two had an unknown outcome.The colonies that contributed the most banded WTSH fledglings recovered from fallout during the course of the study, were Popoiʻa, a mammal-controlled offshore islet colony near a popular beach park, and MCBH-KB, a mammal-controlled onshore colony at a military base with the second highest colony radiance from the study (Table 1; n = 8 fallout events for each colony).However, if the percentage of banded birds recovered (number of birds discovered/number of birds banded) is considered, M anana contributed the most, with over 1.1% of the birds banded recovered following fallout.Fallout birds with bands were discovered from seven of the nine colonies surveyed.

| Spatial and statistical analyses
The mean distance between the natal colony and fallout location was 24.91 km (σ = 27.65;Table 2).The longest distance between a fallout location and a natal colony occurred on a separate island (Maui) from the natal colony (O'ahu), roughly 140 km away.When the outlier of the fallout location on Maui (the only record of interisland fallout in this study) was omitted, the mean distance of the fallout location to the natal colony was 20.50 km (σ = 15.77),conversely, the mean distance of the fallout location to nearest WTSH colony was 9.11 km (σ = 8.74).Only one of the fallout birds analyzed (n = 27) fledged from the colony that was in closest proximity to their fallout location, with 11 fallout locations  (40.80% of the dataset) occurring on opposite coastlines from their natal colony (Figure 3).The mean distance from the fallout location to the natal colony was significantly greater than the distance from the fallout location to the nearest colony (paired t-test t = 5.93, df = 26.00,p = .000003;Figure 4).Fallout location to coastline showed a < 1 km mean distance (x ¯= 0.56 km, σ = 0.68), also with fallout location to a major road or highway (x ¯= 0.71 km, σ = 0.81; Table 2).Both the fallout location to the coastline and fallout location to a major road were significantly shorter distances than the fallout location to natal colony by a measurable magnitude (coastline: paired t-test t = 4.50, df = 24.03,p = .0001;major road: paired t-test t = 4.53, df = 24.09,p = .0001;Figure 5).The fallout location radiance was highest for individuals found in 2021 and 2022, likely due to increases in light pollution island-wide over time, with the average yearly radiance values of fallout location points increasing from 2018 to 2022 (R 2 = 0.49) and a 157% increase in the average radiance of fallout locations compared between 2018 (x ¯= 13.80) and 2022 (x ¯= 35.48;Table 3).The highest radiance value (85.10 nW/sr*cm 2 ) recorded for a fallout location was five times greater than the mean radiance for all fallout locations (x ¯= 14.90 Â 10 À9 W/cm 2 *sr).The average radiance for fallout locations also exhibited higher radiance values compared to the average radiance of the island of O'ahu's habitable areas (x ¯= 6.50 nW/sr*cm 2 ; VIIRS DNB 2018-2022; paired t-test t = 2.25, df = 52.0,p = .0284).

| DISCUSSION
Based on our findings, the assumption that seabird fledglings originate from natal colonies closest to their fallout location was rejected, as distances between fallout locations and natal colonies were significantly higher than expected, especially in comparison to the distance from fallout location to the nearest colony.Of the 27 recovered individuals, only one fledgling originated from the natal colony nearest to their fallout location.Furthermore, more than 40% of fallout fledglings were discovered on opposite coastlines from their natal colony (Figure 3).In one case, a fledgling was recovered postfallout on a different island than their natal colony, more than 100 km away (Figure 3), highlighting how artificial light can pose a risk to fledging seabirds from other islands in a meta-population or island chain.Overall, the findings demonstrate that the relationship between natal colony and fallout location can span large distances and occur in highly varied and difficult-to-predict patterns (Figure 3).Most fallout birds discovered in this study were in close proximity to a coastline, consistent with the geographic location of colonies and the fact that they fledge toward the ocean.However, the high incidence of crosscoastal fallout suggests a potential attraction back to land via coastal lighting after successful fledging (Rodríguez et al., 2015).This suggests fledglings may be attracted back toward artificial lights on the coastline after successful fledging, and/or following the coastline, increasing their risk of fallout from artificial lights regardless of light exposure near the natal colony (Troy et al., 2013).Additionally, light pollution in coastal areas is higher on Oʻahu (and many of the Hawaiian Islands; Figure 2), with greater densities of development for industry, transportation, tourism, and residences oriented along the coast, likely increasing the potential for attraction.Further studies utilizing GPS tracking data could better elucidate potential patterns of returning to coastal areas after successfully fledging.F I G U R E 5 A box plot displaying the distance between the fallout location to the coastline (km) and the fallout location to a major road (km).The inner line indicates the distribution mean with 95% CI.
T A B L E 3 The mean and standard deviation of the radiance (nW/sr*cm 2 ) of fallout locations across years.

Year
x A high proportion of banded birds in this study also experienced fallout near roads, which may be due to the increased presence of artificial lights and electric wires along roadways (Rodríguez et al., 2014;Troy et al., 2013) but could also be due to an increased likelihood of detection along roadways due to heightened anthropogenic activity, and/or potential interaction with the coastline, as urbanization and transport are largely clustered coastally in the Hawaiian Islands.However, one banded individual was recovered at 226 m in elevation, 9.31 km away from the coast, suggesting lights on the interior of the island are also serving as attractants.These results highlight the added importance of managing highway and coastal lighting and orienting recovery efforts in these regions, as these densely-populated areas of high fallout may also increase the likelihood of mortality by automobile collisions, although further research is needed.
Our findings identify a relationship between increased radiance values (indicating light pollution) near natal colonies and at fallout locations.Higher light pollution levels in relation to fledging colonies can be particularly impactful on fallout rates since species in the group Procellariformes are more likely to return to the natal colony to breed compared to other species (Antaky et al., 2021).In recent years, the impacts of artificial light pollution have become more acknowledged and researched, especially as global radiance values continue to increase (S anchez de Miguel et al., 2021).However, scientists are still only beginning to uncover the harmful impacts of light pollution on wildlife and human health, with regulations and mitigation slowly gaining success in implementation.Scientific evidence suggests that artificial light has negative effects on a variety of wildlife including birds, amphibians, mammals, insects, and plants (Marangoni et al., 2022).Coastally the impacts are most felt among seabirds and sea turtles (Thums et al., 2016).Because of this, the benefits of various "dark skies" initiatives such as light reduction and development restrictions, are gaining traction.Designating areas as "dark sky reserves" have proven to preserve the benefits of light-reduced areas for wildlife, human health (Aulsebrook et al., 2018), and night-sky viewing (Zielinska-Dabkowska & Xavia, 2021).
Although there was a relatively small number of recovered bands compared to the number of individuals banded, each banded recovery provided crucial insight into relationships between natal colonies and fallout locations.Three fallout events occurred at sea level at a site with low radiance in a previously identified fallout hotspot in close proximity to the largest WTSH colony in the Hawaiian Islands (DLNR, 2022), but surprisingly, none of the birds at this location came from the large proximate colony (Figure 1, 3; although this location had relatively low banding effort due to a high number of breeding pairs).Instead, fledglings recovered at this hotspot traveled nearly 15 km from smaller colonies before experiencing fallout at this location.This suggests social attraction or other unseen factors may be playing a role in fallout orientations as well, similar to previous findings that indicated some fallout had no discernible patterns or attractants (Troy et al., 2013).
Using banding and GPS tracking data of Cory's Shearwater (Calonectris borealis) fledglings, Rodríguez et al. (2015) found recovered birds at much smaller distances from the natal colony than our study (16 km>).These differences in study results may be due to differences in light orientation, local topography, colony distribution, species-specific behaviors, and so forth, emphasizing the importance of locally-drafted and species-specific fallout management plans based on adaptive management that incorporates the monitoring of banded chicks.The longer distances in our study may be due to increasing levels of urbanization on O'ahu, suggesting that as a region becomes more light-polluted, the variability in fallout may increase as the spatial landscape becomes less predictable (Gaston et al., 2013).Therefore, fallout studies lacking colony origin data should be cautious in assuming fallout individuals came from the nearest colony, as these relationships can be counterintuitive and highly nuanced.
Rejection of our hypothesis could potentially be due to a small sample size, however, the sample size is consistent with many banding and recapture studies.Though the number of recovered bands was a relatively small proportion of all birds banded, these proportions are in line with capture-mark-recapture studies with incidental, citizen science, or passive recapture methodologies (Horswill et al., 2018).Many colonies could not be visited yearly due to weather and access considerations and the proportion of birds banded at each colony varied due to colony size and burrow access.For example, in some colonies, 95-100% of fledglings were banded, and in contrast, <1% of the predicted 30,000+ WTSH that nest on M anana (DLNR, 2022) were accessible, therefore the location and proportion of banding may have skewed the number of birds recovered from each colony.
The exact number of grounded fledglings on the island of Oʻahu was unknown for the study period due to the nature of seabird recovery on the island, where fallout response was not systematic and reporting was dispersed among multiple organizations.Additionally, not all areas of the island were exhaustively searched, and many fallout individuals were likely scavenged prior to discovery or grounded where they were unlikely to be discovered (e.g., mountain ranges, dimly lit areas, isolated areas, or behind visual obstructions).The number of rescued fallout birds on Oʻahu is estimated to range widely each year and has been calculated to be between 74 and 525 individuals (Urmston et al., 2022).Between 25 and 91 fallout birds were detected annually during systematic road surveys conducted in the southeast region of Oʻahu over an eight-year study (Friswold et al., 2020), but the true amount of fallout is likely much greater for both of these values as it is nearly impossible to recover all fallout individuals.Our analyses did not include un-banded fallout, since it was not possible to determine colony of origin.Although a broader assessment of total island fallout warrants further research, we wanted to focus our efforts on understanding relationships between the natal colony to fallout location.Weather and lunar variables were excluded from this analysis due to incomplete weather data and exact fallout dates, although it is possible that weather and lunar cycles played a role in fallout orientation.Lastly, this study likely underestimates the total distance flown prior to fallout, since distances were calculated using a straight aerial route, with the actual distance flown likely farther.
Our results indicate that all artificial lights in a metapopulation or island chain hold some risk to fledging seabirds, with the degree of risk variable and unknown.This suggests that Procellariformes with an IUCN Red List conservation status in the Hawaiian Islands (Tristram's Storm-petrel; Hydrobates tristrami, Band-rumped Stormpetrel; Hydrobates castro, Newell's Shearwater, and Hawaiian Petrel), may be at risk from artificial lights on neighboring islands, even if light reduction measures have been conducted near their source colonies.For instance, in addition to our inter-island fallout recovery, a Band-rumped Storm petrel was recovered following fallout on O'ahu that likely fledged from the island of Maui, based on genetic relatedness (Antaky et al., 2020).Therefore, a large-scale banding effort with an abundant procellariid species, like WTSH, can provide insights into fallout patterns for threatened Hawaiʻi and Pacific species, and further support banding studies and conservation inferences of endangered species with inherent limited sampling capacity.For example, 52 Critically Endangered Newell's Shearwater fledglings were banded at a colony on Kaua'i over five years, and none were later recovered (Ainley et al., 2001) making fallout patterns difficult to study.Our results can assist in informing light pollution impacts on seabirds within a metapopulation and support IUCN Threatened and cryptic species.However, responses in other species may differ, so species-specific behaviors should be considered in management interventions.
In many regions, "lights out programs" and "dark skies initiatives" can be difficult to implement in totality (Crymble et al., 2020;Cuesta-García et al., 2022;Laguna et al., 2014;Rodríguez, Moffett, et al., 2017) due to resistance from federal, state, and industrial entities.Therefore targeted management in identified hotspot areas (i.e., the southeastern shore of O'ahu) are crucial sites to begin light alteration and management with the intention of broad-scale adoption, particularly within five km of seabird colonies (Friswold et al., 2020).Light minimization and reduction methods coupled with rescue and outreach/education programs in hotspot areas during nesting and fledging seasons (specifically peak fallout; Friswold et al., 2020), should be implemented to the range of the metapopulation or island chain to address fallout completely.If it is not possible to fully extinguish lighting; shielding, reorienting, or recoloring lights is likely to reduce light-induced mortality in seabirds (Crymble et al., 2020;Laguna et al., 2014;Moon et al., 2020;Rodrigues et al., 2012;Reed, Sincock & Hailman, 1985;Rodriguez, Dann & Chiaradia, 2017;Rodríguez, Moffett, et al., 2017;Rojek & Region, 2001).
To mitigate fallout, education and cooperation are needed within the transportation, commercial, industrial, military, parks and recreation, tourism, and local community sectors, with efforts coordinated across all of the Hawaiian Islands.Education of residents, particularly those in hotspot areas, may increase the recovery of fallout individuals, increase support for light modification programs, and reduce mortality through community light reduction and recovery.Additionally, cooperative agreements to certify compliance with seabird mortality reduction could be awarded to entities following light alteration recommendations in the Hawaiian Islands.Wildlife managers in Hawai'i have taken multiple steps to address seabird fallout through the creation of dedicated seabird recovery programs, with fallout as a dedicated portion of the focus (MNSRP, Hawai'i Wildlife Center, Save our Shearwaters, and the Kaua'i Endangered Seabird Recovery Project, for example), as well as state and federal support and intervention (DLNR, DOFAW).The island of O'ahu has multiple seabird research and response programs (e.g., PRC and HMAR) however, a state-wide task force across all islands that is responsible for expanded education and fallout reduction and recovery has yet to be developed and is necessary to address fallout in totality.A partnership of this scale is essential to fully protect seabirds and other species from population decline due to light pollution.To achieve this, support and coordination for light management needs to be applied omnidirectionally from the federal and state level to the inter-organizational, community, and individual levels.
Steps to rectify light pollution impacts are crucial in the next decade to prevent further seabird population decline, especially as other seabird mortality factors (sealevel rise, plastics pollution, depleting fish stocks, etc.) continue to intensify and become increasingly complex to manage.Policy changes along with state-wide campaigns in accordance with industrial, commercial, and residential partnerships that put community education, governmental support, and adequate funding for conservation action, are essential to address the breadth and complexity of this issue.The range in distance from fallout to colony exemplifies that all lights in a metapopulation need to be managed, even those seemingly safe from attracting seabirds-requiring stronger communication and cooperation for light reduction and alteration throughout the Hawaiian Islands.Seabirds are declining globally due to a variety of land and sea-based threats (Paleczny et al., 2015) and coordinated efforts to reduce light pollution within a metapopulation or island chain during critical life history stages are necessary to increase survival and contribute to global seabird recovery.
Oʻahu banding activities and project management were led by the Hawaiʻi Wildlife Ecology Lab in the Natural Resources and Environmental Management Department at the University of Hawaiʻi at M anoa in collaboration with the State of Hawaiʻi Department of Land and Natural Resources (DLNR), and Division of Forestry and Wildlife (DOFAW) among other project partners.Some WTSH banding took place during DLNR and DOFAW's yearly breeding pair counts at certain banding locations.Collaborating organizations for training, banding, and collaboration include Maui Nui Seabird Recovery Project (MNSRP), Pacific Rim Conservation (PRC), MCBH-KB, and Hawai'i Marine Animal Response (HMAR).

F
I G U R E 1 Locations of wedge-tailed shearwater (WTSH) breeding colonies on the island of Oʻahu, Hawai'i.The closed circles indicate where WTSH banding was conducted for the years 2018-2022 (n = 9).

F
I G U R E 2 Fallout locations on O'ahu, Hawai'i (n = 27) with a radiance layer indicating increasing radiance intensity by color using 2022 VIIRS NOAA data (lightpollutionmap.info).T A B L E 1 Banding data across all years and surveyed colonies.

FFallout
I G U R E 3 Dotted lines indicate the flight path from natal colonies (indicated as yellow markers; n = 9) to the fallout location (indicated as open circles; n = 27) on O'ahu, Hawai'i.Routes are displayed as direct linear paths between where banding was conducted pre-fledging to the reported location of the recovered fallout.Locations of un-surveyed colonies are indicated as white markers; n = 9.The map was created using ArcGIS (ArcGIS online, 2022).MCBH-KB = Marine Corps Base Hawaii Kaneohe Bay, KBP = Kailua Beach Park.The upper right corner shows the inter-island fallout event discovered on Maui of a bird banded at a colony on O'ahu.T A B L E 2 Recovered band data depicting calculated direct aerial route distances and fallout location characteristics (n = 27) for years2018-2022.1.24-31.50Note: x ¯= mean, range = the minimum and maximum value in the dataset, SD = standard deviation.Distance calculations were taken from center points across fallout locations and colonies which can vary up to 0.90 km.

F
I G U R E 4 (a) Histogram of distance in km from fallout location to natal colony (n = 27) for recovered band data.The dashed line indicates the mean value (x).(b): Histogram of distance from fallout location to the nearest colony in km (n = 27).