Final report for OW18-017
Hawai‘i’s increased agricultural biodiversity has attracted an influx of non-native and introduced avian and rodent pests that cause an estimated total of over $100 billion in damage across the United States, with farms statewide reporting up to 80-100% of crop loss due to these pests. Introduced avian and rodent species also act as vectors for disease, dispersal agents for noxious weeds, and competitors with native species. Pest deterrent techniques are costly, impermanent, and often ineffective. Raptors, especially owls, have proven an effective form of biocontrol. Pueo, or the Hawaiian short-eared owl, is the only endemic avian predator on O‘ahu and Maui, with pellet content demonstrating a dietary inclination toward non-native avian, rodent, and invertebrate species. We used a standardized survey protocol to assess: the distribution and abundance of Pueo on O‘ahu and Maui agricultural lands; determine Pueo seasonal habitat use of agricultural lands, through tagging; examine owl pellets to assess potential reduction in crop predation by pest species; and produce a list of possible measures to increase Pueo abundance on agricultural lands, to be applied by farmers and landowners. Pueo occupancy was found to be most influenced by prey biomass distribution and elevation. Diet was found to be consistent with previous studies, and likely is proportional to availability. Further, pueo were found to have much higher site fidelity than continental Short-eared Owls, suggesting that pueo on agricultural lands are occupying the same sites year-round, both providing benefits to farmers year-round, and receiving impacts from potential threats such as secondary poisoning from rodenticide or drowning in water troughs. Through farmer-to-farmer and researcher-to-farmer interactions, we engaged producers, discussed their knowledge of pueo, and shared our results. Producers are now helping to achieve a “win-win-win” for the native Pueo, for Hawai‘i conservation, and for economic benefits to agriculture.
1. Assess the distribution and abundance of Pueo on O‘ahu and Maui, including on agricultural lands.
2. Evaluate potential reduction in pest species through diet analysis on owl pellets and surveys.
3. Determine seasonal habitat use by Pueo.
4. Develop Habitat Conservation Recommendations for Producers.
5. Increase producer and public awareness of the role of Pueo in agriculture.
- (Educator and Researcher)
Methods (Organized by Project Objectives):
- Assess the distribution and abundance of Pueo on Oahu and Maui, including agricultural lands.
Study area and site selection
The island of Oʻahu, the third largest island in the Hawaiian archipelago, is home to roughly two-thirds of the human population of the state of Hawaiʻi. Including small offshore islets, Oʻahu is 1,545 km2, making it the 20th largest island in the United States (State of Hawaii 2019a). The island of Maui consists of 1,883 square km (188,300 ha) of land and is the second largest island in the Hawaiian Island archipelago (Fig. 2) (Sinton, 1987). Major land cover types include cropland, shrubland, grassland, forest, and developed lands (see Appendix A) (ArcGIS Pro, version 2.5.1, Esri). Cropland habitat at the time of surveys was in the process of transitioning from fallow sugarcane to diverse croplands, and experienced a frequent and severe fire season in 2019 that left some areas unusually dry and barren (pers. obs.; pers. comm. with ag. producers).
Focal surveys. Focal surveys were carried out at six selected sites on Oahu: (1) Dole Plantation (Dole; 21°32’N, 158° 02’W), which consists of pineapple crop and fallow land interspersed with tracts of California grass (Brachiaria mutica) and alien forest; (2) Hamakua Marsh (Hamakua; 21°23’N, 157°44’W), a wetland bordered by shrubland and development; (3) Kawainui Marsh (Kawainui; 21°23’N, 157°46’W), an expansive wetland fringed by alien forest; (4) U.S. Navy Lualualei VLF Transmitter (Lualualei; 21°25’N, 158° 08’W), an expansive grassland encircled by alien forest and development; (5) Marine Corps Base Hawaii- Kāneʻohe Bay (MCBH-KB; 21°26’N, 157°45’W), a series of wetlands fringed by shrubland; and (6) University of Hawaiʻi West Oʻahu (UHWO; 21°21’N, 158° 03’W), a mosaic of agricultural fields, shrubland, and scattered grasslands. Daily activity surveys were conducted at four of six sites (Dole, Hamakua, Lualualei, and MCBH-KB).
We conducted a total of 20 focal surveys to determine daily patterns of behavior between April 20, 2018 and June 17, 2018. Individual owls were observed and behavioral observations were recorded to the nearest minute. Behaviors were categorized as soaring, hunting, wing-clapping, or perching. Observation times were standardized as minutes after civil twilight in the mornings and minutes until twilight in the evenings. Barn Owl activity was incidentally recorded.
To collect information on breeding activities, we conducted 101 surveys at six sites on Oahu between December 30, 2017 and August 19, 2018. Breeding activity surveys started 60 minutes before sunset and ended at civil twilight. In addition, we included breeding activity observations from the daily activity surveys, and island-wide population surveys.
Island-wide population surveys. Locations for population surveys were determined using a random stratified design across six vegetation types: wetland, agricultural land, grassland, shrubland, alien forest, and native vegetation. Thirty-five survey sites were randomly selected from the six vegetation types using ArcGIS Pro. Surveys were carried out from vantage viewpoints, defined as a point within 500 m of the randomly selected sites with at least a 180° field of view of the surrounding area, extending at least 300 m from the viewpoint. Surveys were carried out in the late evening, between 75 and 60 minutes before sunset, and finished at civil twilight (when the sun is 6° below the horizon), typically 30 minutes after sunset. Surveys were repeated three times at each site to increase the likelihood of detecting Pueo, and were spaced, when possible, at least three weeks apart.
Citizen science and web reporting
Citizen science observations were gathered through the Pueo Project online portal (www.pueoproject.com), a website created for this project in 2017, and eBird (www.ebird.org), a globally recognized reporting system for bird sightings. Pueo sightings were reported to the Pueo Project website by filling in an online form or by providing observation information into an embedded ESRI GIS web app. Sightings were gathered from March 15, 2017 to March 15, 2018. All available eBird data on Pueo sightings in Hawaiʻi were downloaded on March 15, 2018. By overlaying a GIS layer of data points with the vegetation layer, we identified the vegetation type where each observation took place.
Focal surveys. Relative Pueo activity was estimated as the number of Pueo observations per 30-minute interval. Time intervals were standardized to indicate time since civil twilight in the mornings or time until civil twilight in the evenings. We used a t-test to compare mean time between twilight and Pueo observations between morning and evening.
Island-wide population surveys. Pueo densities were calculated for each vegetation type by dividing the maximum number of Pueo observed in a single survey for each site by the total area (ha) surveyed for that vegetation type. The Pueo population size on Oʻahu was estimated by multiplying the density of Pueo per vegetation type by the total area of that vegetation type available on Oʻahu. Total areas of each vegetation type were calculated with ArcGIS Pro 2.1.1., based on the GIS vegetation layers provided by the State of Hawaii’s Office of Planning (State of Hawaii 2019b). Confidence limits were obtained by bootstrapping in IBM SPSS Statistics v.25. The confidence limits were derived by adding the 999 unsorted bootstrapped estimates for the populations, sorting them into numerical order and taking the 25th and 975th values. Using an independent samples t-test, densities of Short-eared Owls were compared between published studies that utilized targeted approaches, in which known populations or breeding sites were surveyed, and studies using non-targeted approaches, which utilized random site selection techniques. All statistical analyses were carried out using IBM SPSS Statistics v.25.
- Evaluate potential reduction in pest species through diet analysis on owl pellets and island-wide surveys of potential prey species.
Pueo pellets were collected on Oahu and Maui, and were dissected and analyzed. Birds, small mammals, and insects (Orthoptera) were also surveyed.
Site Selection. Access to approximately 70 percent of the island of Maui was granted by cooperating landowners and managers. Potential survey sites were located by rasterizing the island of Maui into 500 by 500-m units using R packages ‘raster’ (Hijmans et al., 2017) and ‘rgdal’ (Bivand et al., 2018) in the statistical programming software R (version 3.5.2) (R Core Team, 2020; RStudio Team, 2020). To ensure sampling across an elevational gradient and multiple vegetation types, potential survey areas were divided equally into three altitudinal bands (0-1000 m, 1000-2000 m and >2000 m). An equal number (n=15 per band) of survey sites was randomly drawn from these three altitudinal bands. Survey sites were evenly sampled across all dominant vegetation types and adjusted within a 500 m radius if accessibility was limited due to terrain.
Surveys. At each site, surveys took place over three days and two nights. The dominant vegetation type within 200 m surrounding the survey point was determined using the ‘Buffer’ tool in ArcGIS Pro (version 2.5.1, Esri). Site characteristics (ground cover, canopy cover, and height of dominant vegetation) were collected upon arrival at the survey point from three randomly selected one-meter square quadrats (Miller et al., 2017). A clinometer or measuring tape was used to estimate vegetation height and a densiometer was used to estimate canopy cover (James & Lockerd, 1986). Using a field thermometer, mean temperature per survey was determined by the temperature at the start and end of each audio/visual avian survey. Mean annual rainfall at each survey point was extracted from the online Rainfall Atlas of Hawai‘i database (Giambelluca et al., 2013). At each survey point, three different wildlife survey methods were utilized to evaluate the abundance of birds, bats, rodents, and insects (Orthoptera). Pueo pellets were collected opportunistically.
Bird and bat point count distance surveys. Bird and bat audiovisual surveys started one hour prior to sunset and ended thirty minutes after sunset (Cotin & Price, 2018). During the duration of each 90-minute survey, all aerial vertebrate species (bat, bird) seen or heard up to 200 m from the survey point were recorded. At the end of every audiovisual survey period, two distinct short-eared owl vocalizations from the Merlin Bird application were broadcast from a small Bose speaker twice, thirty seconds apart. The speaker was positioned between one and two m above the ground with the volume set for human ears to hear at a maximum distance of 100 m (Boscolo et al., 2006).
Rodent surveys. Capture, mark, and recapture were utilized to target rodents on the landscape, the house mouse (Mus musculus) and three rat species (Rattus exulans, Rattus norvegicus, Rattus rattus), using 50 Sherman traps (standard model LFA-TDG, 7.5 x 9 x 23 cm), baited with rolled oats, and set five meters apart in a 2 x 25 grid formation to estimate population density of a given vegetation type (area= 625 m2 ) (Hoffman et al., 2010; Shiels, 2010). Traps were set prior to sunset and checked the following morning. All traps were re-baited after the first night of trapping and removed after two nights for a total of 100 trap nights per survey point. Two nights of trapping (100 trap nights) have been determined sufficient for providing an accurate index of rodent abundance (Krebs, 1966; Thirgood, 2003). Trapped rodents were anesthetized, biomass was recorded, and rodents were marked on both ears and tail with a black marker. After reviving from anesthesia, the rodents were released back into the environment.
Insect (Orthoptera) sweep net surveys. A combination of high (~20 cm), fast (0.9 s), low (~5 cm), and slow (1.3 s) sweeps were performed using a medium-size net (0.4 m in diameter and 0.6 m in depth) to sample a 10 x 50-m area (area= 500 m2) at each survey point for ten minutes approximately an hour prior to sunset (Larson et al., 1999). Captured individuals were photographed, their length (head to abdomen) was recorded, then released back into the environment (Eklöf et al., 2017; Orinda et al., 2017).
Determination of biomass. The biomass of bird and bat species detected during surveys was determined by searching the Birds of North America and the Encyclopedia of Life online databases (The Cornell Lab of Ornithology’s guide to birds of North America; 2001; Encyclopedia of Life). For analysis, bird species were grouped into two size classes: “Small” (0-30 g, e.g. house finch, Carpodacus mexicanus) and “Large” (30-350 g, e.g. zebra dove, Geopelia striata, or grey francolin, Francolinus podcicerianus). Pueo are not known to consume prey items greater than their own body mass, so the maximum biomass of a potential prey species predated by a pueo was considered anything lower than the mean body mass of a pueo (350 g) (Wiggins et al., 2006; Mostello, 1996; Mostello & Conant, 2018).
The biomass of rodents was derived by weighing individuals. The biomass of insects (order Orthoptera) was derived by measuring the length and classifying individuals (Eklöf et al., 2017; Orinda et al., 2017; Encyclopedia of Life). Borror and DeLong’s invertebrate dichotomous key was utilized for species identification (Tripplehorn & Johnson, 2005; Eklöf et al., 2017; Orinda et al., 2017; Encyclopedia of Life).
Authorizations. All activities were conducted following Institutional Animal Care and Use Committee permit: #3094, Institutional Review Board permit: #2019-00064, Hawai‘i Department of Land and Natural Resources (DLNR)- Division of Forestry and Wildlife permit: MDF-060319A, DLNR- Natural Area Reserve and Native Invertebrate Research permit: l1333, and Haleakalā National Park federal research permit: HALE-2019-SCI-0001.
Prey bird distance models. Multi-covariate detection functions and density surface models were produced for the “small” and “large” prey bird size classes to predict species abundance using the ‘dsm’ and ‘Distance’ packages and ‘predict.glm’ function in the statistical programming software R (Miller et al., 2019; R Core Team, 2020; RStudio Team, 2020). Biomass was derived from the predicted abundance outputs from the top distance sampling models for the “small” and “large” bird size classes.
Rodent and insect (Order Orthoptera) generalized linear models. Generalized linear regression models were created in R to predict mouse, rat, and insect (Orthoptera) relative abundance and relative biomass (R Core Team, 2020; RStudio Team, 2020). All models were evaluated for appropriateness by examining diagnostic plots. Models explaining the most deviance based on the ANOVA F-test results in R were retained for inference (Burnham & Anderson, 2002; R Core Team, 2020; RStudio Team, 2020). The coefficient of determination (r2) for each model was derived using the ‘modEvA’ package and ‘RsqGLM’ function in R (Barbosa et al., 2015; R Core Team, 2020; RStudio Team, 2020).
Pueo occupancy. Occupancy models were constructed with vegetation characteristics and predicted index of prey biomasses using the package ‘unmarked’ in R (Fisk & Chandler, 2011; R Studio Team, 2016; R Core Team, 2018). Predicted bird densities were extracted using the ‘predict.glm’ function in R and an index of densities were extracted from rodent and insect (Orthoptera) GLM’s (R Core Team, 2020; RStudio Team, 2020). Graphs were created using the ‘ggplot2’ package in R and derived using the mean value from each vegetation type (Wickham, 2016; R Core Team, 2020; RStudio Team, 2020).
Multiple single-season occupancy models were run using the package ‘unmarked’ to examine which covariates influence pueo detectability and occupancy (Fisk & Chandler, 2011; R Core Team, 2020; RStudio Team, 2020). Candidate models were ranked by AIC using the package ‘AICcmodavg’ and the top model was retained for inference (Fiske & Chandler, 2011; Kery & Royle, 2016; Mazerolle, 2019). Occupancy models jointly model the ecological process of species occurrence (psi) and the observation process (p) of species detection but estimate these as separate processes (MacKenzie et al., 2017). Detection covariates are utilized to account for the imperfect observation process of species detection data to more accurately predict true occupancy states (MacKenzie et al., 2017). Detection covariates included vegetation characteristics (e.g. canopy cover, ground cover, vegetation height). Occupancy covariates included vegetation characteristics (e.g. canopy cover, ground cover, vegetation height), elevation, mean annual rainfall, and the predicted prey biomass (bird, rodent, and insect (Orthoptera)) data. The ‘cor.test’ function in R was utilized to identify potentially autocorrelated covariates (R Core Team, 2020; RStudio Team, 2020). As a result, temperature was removed from downstream analyses due to autocorrelation with elevation (Pearson’s r = -0.95, p<0.001). I explored the suite of detection variables and retained the model with the lowest Akaike’s information criterion (AIC; Akaike, 1974) for inference (MacKenzie et al., 2017). Bootstrapping and a Pearson’s Chi-squared goodness-of-fit test was utilized for model evaluation (Royle & Dorazio, 2008; Fiske & Chandler, 2011; MacKenzie et al., 2017; R Core Team, 2020; RStudio Team, 2020).
- Determine seasonal habitat use by Pueo.
Specific trapping sites on O‘ahu and Maui were selected in areas where preliminary surveys identified frequent pueo use and land access was granted (Cotin and Price 2018, Luther 2020). Despite substantial effort and consistently observing pueo, no pueo were captured from the central O‘ahu or Maui field sites; all captures occurred on the leeward side of O‘ahu.
Capture Methods. Pueo trapping sessions were conducted during crepuscular periods known to have peak activity, starting about 2 hours before sunset and ending before midnight from February 2019 through October 2019 (Cotin et al. 2018). Pueo were captured using a dome shaped bal-chatri trap baited with a mouse (Mus musculus; Bird and Bildstein 2007). Captured pueo were outfitted with an alpha-numeric colored visual identification band and a United States Geological Survey aluminum Federal Bird Band (3.2g total). Pueo were outfitted with a VHF transmitter (American Wildlife Enterprises 8.7g backpack AWE-Q) using a backpack-style harness constructed with 3/16-inch Teflon ribbon (Bally Ribbon Mills, 2g).
For each captured individual, several biometric measurements were collected using a wing rule and calipers accurate to the nearest 1 mm including unflattened wing chord, tail, and metatarsal length, and body mass to the nearest 0.1 g using an electronic scale (Appendix B). Individuals were sexed and aged by plumage characteristics (Martínez-Climent et al. 2002). All activities took place under appropriate federal, state, and institutional permits (Bird Banding Lab permit no. 24137; Hawai‘i Department of Forestry and Wildlife Scientific Collecting Permit no. WL19-10; University of Hawai‘i Institutional Animal Care and Use Committee protocol no. 18-2752; and University of Hawai‘i Institutional Biosafety Committee protocol no. 18-11-949-01).
Tracking Methods. VHF tagged pueo were resighted using a 3-element yagi antenna on at least three different days per week over the lifespan of the transmitters. Sequential resight locations were collected at least half an hour apart. Upon resighting, we recorded the location of the tagged pueo, and conducted behavioral observations for 5-10 minutes. We observed the individual with a spotting scope or binoculars either before they were alerted to our presence or from the greatest distance possible to minimize behavioral response to observer presence. Behavior at the time of resighting was classified as either flying, hooting, hunting, patrolling, agonostic, wing-clapping, courtship flight, roosting, or unknown. Additional data were collected during each resight including time of day, general habitat notes (e.g. perch substrate), visually estimated perch height, and weather data. Weather metrics included temperature and windspeed measured with a Kestrel 5000 Environmental Meter, precipitation (none, slight, steady, or heavy), and percent cloud cover.
Resight data were analyzed with the program R (Version 4.0.3) using the packages adehabitatHR, sp, and rgeos, to estimate pueo home ranges and describe habitat use (Calenge 2006, Bivand et al. 2013, Bivand and Rundel 2020, R Core Team 2020). To assess differences in habitat use by behavior, we used a Pearson’s Chi-square test to compare land-cover class use while hunting and while roosting for all individuals with >40 resights pooled. Low count categories were lumped or excluded, or a Fisher’s exact test was used when the data failed to meet the minimum expected count assumptions of a Pearson’s Chi-square test (Kim 2017). A Pearson’s Chi-square test was used to compare activity and time period for all individuals with >40 resights pooled, and in the case of low counts, a Fisher’s exact test was used. If results from a pooled Chi-square or Fisher’s exact tests were significant, a Fisher’s exact post-hoc analysis was conducted (Shan and Gerstenberger 2017).
- Develop Habitat Conservation Recommendations for Producers.
Best management practices were developed through discussion with cooperating producers, as follows:
1) Habitat management: mowing length recommended, leave naturalized edge areas where possible, perch sites (fencing, trees)
2) Nest site awareness: preferred nest-site characteristics, behaviors; as ground nesters, pueo are extremely vulnerable to trampling and predators. Watch for courtship and nesting behaviors, food provisioning, and owl pellets
3) Prey requirements: numbers of prey items (insects, birds, rodents) removed by nesting and non-nesting owls
4) Rodenticide avoidance: use of A24 “stun” traps allow non-toxic rodent kill, with safe prey item provided for opportunistic owls, and no poison moving up the food chain
- Increase producer and public awareness of the role of Pueo in agriculture.
Maui Partners. We visited Maui stakeholders beginning in October 2018. The initial Maui stakeholders included Alexander & Baldwin, Haleakala Ranch, and East Maui Watershed Project. We visited each stakeholder at their property and took a tour of the area. We discussed the needs and expectations of the partnership, pest issues, and the survey plan. We provided materials for employees to record any observations to understand what areas of the property pueo are seen. Greg Friel and Jordan Jokiel of Haleakala Ranch and Dan Eisenberg of East Maui Watershed Partnership agreed to provide land access and technical support. Alexander & Baldwin was in the process of shutting down their sugarcane operations but still provided support in principle for the project.
In November 2018, the Pueo Team had an educational booth, along with other wildlife organizations at the first Hawaiian Bird Symphony Event (~1500 attendees). This event, spearheaded by Dr. Price, included performance of original compositions and visual art projects based on songs of native Hawaiian birds, both extinct and extant, including pueo. We had a museum specimen of pueo and talons to show children and provided material regarding pueo identification and current research efforts at the booth. Further Hawaiian Bird Symphony Events in other locations brought in an additional 4500 attendees.
The Pueo Team (Dr. Price, Dr. Cotin, Prof. Steensma, Chad Wilhite, and Laura Luther) met with Michelle Starke, Bruce Schnicker, and Yarrow Flower, managers from Bayer (formerly Monsanto), to discuss a partnership to access their lands to monitor pueo and provide feedback on habitat preferences and the feasibility of pueo as a biocontrol. We visited Bayer’s main property on O’ahu in October 2018 for a tour of the landscape to identify potential pueo habitat and vantage points to survey from. We left material with Bayer employees to report any observations. In Spring 2019, we met with Bayer employees to provide an informational training about pueo observation and reporting during a ‘Lunch and Learn’ session.
Members of the Pueo Team also met with Darren Strand of Mahi Pono, along with Mae Nakahata (formerly of Alexander & Baldwin), to discuss a partnership to access their lands, in February 2019. Mahi Pono is one of the largest agricultural entities on Maui, having begun diversification of agriculture after sugarcane production ended on the island. They were interested in surveying for pueo and learning about the potential of pueo as biocontrol. Pueo Team members also met with Kristin Mack Almasin of Ulupalakua Ranch who agreed to provide land access and technical support.
All potential field sites were re-visited in February of 2019 to follow up with stakeholders to answer any questions, and discuss project implementation. Discussions with ranch managers and farm managers were then ongoing, as the intensive field season of data collection occurred from May 2019- February 2020. Ideas for practical management efforts were exchanged as the season progressed.
In February of 2020, right before the covid-shutdown of travel among islands, our team held workshops on Maui with stakeholders to report results, share materials, discuss outcomes, and listen to feedback. We presented talks to the Haleakala Ranch team at their office, to a group of ranchers and Haleakala National Park employees at Ulupalakua Ranch, to members of the Maui Conservation District and Maui Farm Bureau, and to the Haleakala Chapter of the Hawaii Farmers Union, to promote sustainable practices related to owls with a broad cross-section of producers over several days in February 2020.
Throughout the project we were in communication with non-producer land managers across the Hawaiian Islands to discuss threats to pueo and historic mortality data: Kauai Seabird Recovery Project, Maui Nui Seabird Recovery Project, Hawai’i Wildlife Center, and the Department of Transportation.
We were also in communication with interested volunteers and farmers through the Pueo Project website (www.pueoproject.com). We received useful reports on pueo activity across the Hawaiian archipelago. Our lab made a connection with Brynn Foster, owner of Voyaging Foods Farm on O’ahu. She has observed pueo and is interested in the Pueo Team visiting her land to recommend habitat preferences of pueo. We also spoke with Bronwyn from Molokai’s farm sanctuary, Hui Ho’olana; she wants to promote pueo habitat on the sanctuary and would like the Pueo Team to visit and make further recommendations.
Results (Organized by Project Objectives):
1. Assess the distribution and abundance of Pueo on Oahu, including agricultural lands.
A total of 21 Pueo detections were recorded during morning surveys and 32 during evening surveys. Most detections occurred within an hour of civil twilight. The distribution of detections did not significantly differ between morning (mean = 91 ± 107 minutes) and evening (mean = 72 ± 72 minutes; t51 = 0.792, p = 0.43). The mean number of observations per survey did not differ between mornings (2.1 ± 1.2 observations) and evenings (3.2 ± 1.7 observations; t18 = -1.530, p = 0.14).
Number of Pueo observations during surveys varied among sites, with only one of the three studied wetlands lacking any observations (Table 1). Courtship displays were observed as early as November 2 and as late as June 10. Barn Owls were observed during multiple surveys, occasionally at the same time and area as Pueo. Also we gathered information from eight nests on three different islands: O’ahu (n = 5), Maui (n = 2) and Kauai (n = 1). Nests were found in spring (April through June) and autumn (November), and from sea-level up to 2000 meters elevation. Number of eggs ranged from 3-7 per nest. Vegetation types included wetlands, open grasslands and wet montane forests (uluhe fern, Dicranopteris linearis, understory with ʻōhiʻa trees, Metrosideros polymorpha, as primary canopy cover).
Distribution surveys, population densities and estimates on Oʻahu
We carried out 105 surveys at 35 sites, across a total of 1030 hectares (0.66% of the total area of Oʻahu). We detected Pueo at six different sites out of thirty-five, including one wetland site, two agricultural sites, one grassland site, one shrubland site, and one native vegetation site. Densities were highest in agricultural lands (3.3 Pueo per 100 ha), followed by native vegetation (1.0), wetland areas (0.7) and shrublands (0.6). Two vegetation types, grasslands (0.3) and alien forest (0.0), had the lowest densities of detected Pueo (see Table 2). Pueo were observed during survey hours in mowed grasslands, and outside survey time at short grasslands, but never in sites classified as tall grasslands. Based on observed densities, the number of Pueo inhabiting Oʻahu was estimated at 807 individuals (95% CI 8-2199; table 2).
Densities of Pueo on Oʻahu [0.645 (0.006 – 1.756 Pueo per ha)] were in the mid-range, compared to Short-eared Owl studies globally. Studies in Europe targeting known breeding sites for the species reported significantly higher densities (1.8 ± 1.81 owl/100 ha; p = 0.049, t 6.378 = -2.427) than studies in North America which surveyed all of possible habitats and included areas not known to be occupied by owls (0.15 ± 0.27 owls/100 ha).
Fifty-five citizen science reports of Pueo sightings were submitted, during a one year period (March 15, 2017 – March 15, 2018). The highest number of reports of Pueo detections were in developed areas (22%), followed by alien forest (20%), grasslands (20%), agricultural lands (16%), and shrubland (11%). The lowest number of reports were from wetlands (7%) and native vegetation (4%). Ten additional reports submitted as Pueo were identified as Barn Owls after further descriptions were elicited from the citizen scientists. From the eBird portal, 43 reports were obtained across 32 years (1986-2018) for Oʻahu. The first recorded sighting was in 1986, with 50% of the sightings occurring from 2011-2015. Only four reports were from 2017 and 2018. Developed areas had the highest number of eBird Pueo reports (37%), followed by wetlands (26%), and shrubland (16%). The lowest number of sightings were in alien forest and native vegetation (7%), grasslands (5%) and agricultural lands (2%). The highest number of observations were in developed areas for the Pueo portal and eBird datasets, but in agricultural areas for standardized surveys. Conversely, vegetation types with the lowest number of observations were alien forest for the standardized surveys, native vegetation for Pueo portal, and agricultural lands for eBird data.
Based on our surveys, the Hawaiian Short-eared Owl is a crepuscular-nocturnal species, with an elastic breeding phenology that extends more than half the year, November through June, coinciding with the nesting seasons of many waterbirds, forest birds, and seabirds in the islands. Densities on Oʻahu were similar to densities identified in continental, temperate regions utilizing similar non-targeted, randomized and standardized survey approaches (Lehman et al. 1998, Miller et al. 2016, 2017). Similar to populations elsewhere, Pueo may occupy territories with high prey availability, but leave unoccupied low-prey-density territories, even if they contain adequate vegetation. Pueo appear to exhibit flexibility in daily behavioral cycles on the other Hawaiian Islands. While our daily activity surveys suggest they are crepuscular in lowlands on Oʻahu, reports from other islands in the archipelago (such as Lānaʻi, Kauaʻi or Hawaiʻi) suggest that Pueo are more diurnal at cooler, higher elevations (Fern Duvall, pers. comm.). Further study is needed to compare behavioral patterns among islands.
Like the Galapagos Short-eared Owl (De Groot 1983), Pueo nests and owlets have been discovered during virtually all months of the year (Berger 1981). Because we observed courtship displays as early as November (Cotín et al. 2018), well outside the pair formation stage in North America (Clark 1975, Wiggins et al. 2006), we speculate that the Pueo has a more elastic breeding season than its continental counterparts.
Pueo sightings obtained through citizen science efforts were highly biased toward developed and urbanized areas. This is consistent with expected biases of citizen science datasets (Robinson et al. 2018), and is likely an artifact of the data source, since developed and urbanized areas are not likely the most suitable habitat for the species due to high rates of mortality from car strikes (Booms et al. 2014), and invasive predators. Since human populations are most dense in urban areas, the abundance of people increases the likelihood of detecting Pueo that pass through these areas. On Oʻahu, urban areas are often adjacent to suitable foraging habitat, such as urban wetlands, agricultural lands, or open parks, so residents are likely to observe Pueo as they pass through while utilizing nearby habitats.
Pueo were mostly detected in open vegetation types. Agricultural lands had higher densities of Pueo, but nesting activities detected on wetlands indicate the importance of this habitat for the species. Densities obtained with our surveys are aligned with those studies targeting known Short-eared Owl populations with a high rate of occupancy (Goddard 1935, Lockie 1955, Village 1987, Shaw 1995). Densities on Oʻahu are similar to the ones in non-targeted, randomized and standardized studies carried out in Idaho and Utah (Lehman et al. 1998, Miller et al. 2016, 2017), where owls occupy territories with high prey availability, but leave unoccupied low-prey-density territories, even if they contain adequate vegetation requirements.
Despite the small proportion of wetlands on Oʻahu (around 1% of the total surface), all datasets show a relatively high proportion of Pueo observations in wetland vegetation. This could be explained by: (1) owls favoring this kind of habitat, as they do in other parts of the world (Booms et al. 2010); (2) predator control programs in these areas, which lower the risk of predation on eggs, chicks, and adults; (3) and the regular occurrence of waterbird surveys throughout the year, which increases the number of birdwatchers in these areas and the likelihood of detection. Additionally, Pueo nests have been reported in this habitat on the Hawaiian Islands, indicating the importance of wetlands for this species. In contrast, there was a low percentage of observations in alien forests, despite the fact that this vegetation type covers 35% of O‘ahu. Owl detectability is expected to be lower in closed canopy vegetation, even if they are present, and this might provide an alternative explanation for the lower number of observations.
Overall, in our surveys, Pueo were most often detected in open vegetation types, including agricultural lands, grasslands, and wetlands. Pueo densities in native vegetation, wetlands and shrubland were above 0.5 Pueo per 100 hectares. Surprisingly, grasslands, which are the preferred habitat of this species in other parts of the world (Booms et al. 2014), had lower densities than other vegetation types. When divided into subtypes, densities were closer to expected values in grasslands of low height. Invasive California grass (Brachiaria mutica) can grow up to 2.5 meters (8 feet), and sugar cane (Saccharum sp.) up to 6 meters (20 feet), creating fields that are probably too dense or tall for breeding or foraging Pueo, and reducing detectability during surveys. Further study is needed regarding prey availability, predator risk, and competition with Barn Owls or other invasive animals to elucidate why distribution among vegetation types differs from continental systems.
We surveyed ~1,000 hectares on Oʻahu, which is less than 1% of the total area of the island. This relatively small area might not be representative of the whole island, and resulted in a high degree of uncertainty in our estimated population size for Oʻahu (i.e. a large range between low and high estimates). Several factors limited survey effort and should be addressed in future studies. Due to very low detection rates, survey sites were visited three times in order to accurately assess the presence of the species.
Information gathered through the citizen science portal is highly valuable for obtaining phenology and breeding event observations (nests, owlet locations, display flights) or to give insight into areas or sites not accessible to the researchers or the general public. However, data collected in this manner tend to be biased due to the lack of standard distribution of the observers, which hampers their usefulness for running distribution models or other population analyses.
2. Evaluate potential reduction in pest species through diet analysis on owl pellets and island-wide surveys of potential prey species.
Sites. Twenty-six sites were accessed out of the 45 potential sites that were randomly identified. Sites were often a mosaic of multiple vegetation types and characteristics, but the major land cover type and characteristics were recorded where the survey methods took place.
Prey bird detection probability models. Prey bird abundance (14 ± 5 individuals per 1,000 m2) and biomass (931 ± 461 g per 1,000 m2) was highest in low elevation, low to medium vegetation height, and open cropland areas, and consisted of primarily non-native species. Most of the “large” birds observed in lower elevation areas were spotted doves (Streptopelia chinensis). They were observed foraging in croplands during the day then flew towards human development and nearby wetlands to roost at night. Occurrence was also higher near water sources, such as cattle troughs or active croplands with irrigation. Prey bird abundance (4 ± 0 individuals per 1,000 m2) and biomass (148 ± 9 g per 1,000 m2) was lowest in high elevation, shrubby areas which contained both native and non-native species, but native species (including bats) were typically observed above 1,200 m elevation. (Appendix D).
Relative rodent abundance and biomass models. Forested areas with high canopy cover contained the highest relative rodent abundance (6 ± 8 individuals per 1,000 m2) and biomass (380 ± 646 g per 1,000 m2). High elevation, shrubby areas contained the lowest relative rodent abundance (3 ± 1 individuals per 1,000 m2) and low elevation cropland areas contained the lowest rodent biomass (43 ± 25 g per 1,000 m2). Mouse occurrence was moderate to high across all vegetation characteristics especially in short vegetation with low canopy cover (5 ± 3 individuals per 1,000 m2), while rats occurred most frequently in areas with high canopy cover and medium to tall vegetation height (2 ± 2 individuals per 1,000 m2). Across sites the average biomass of a single mouse was 11 ± 2 g and the average biomass of a single rat was 85 ± 17 g. (Appendix E).
Relative insect (Orthoptera) abundance and biomass models. Areas with high canopy cover, high ground cover, and tall vegetation height contained the highest relative insect (Orthoptera) abundance (46 ± 57 individuals per 1,000 m2) and biomass (13.4 ± 14.8 g per 1,000 m2). Low elevation, cropland areas contained the lowest relative insect (Orthoptera) abundance (13 ± 3 individual per 1,000 m2) and relative biomass (3.1 ± 0.2 g per 1,000 m2). Species that were observed were primarily non-native generalists and included the common conehead (Neoconocephalus spp.), field cricket (Gryllus spp.), house cricket (Acheta domesticus) katydid (Microcentrum rhombifolium), red-legged grasshopper (Melanoplus femurrubrum), spur-throated grasshopper (Melanoplus ponderosus), and two-striped grasshopper (Melanoplus bivittatus). (Appendix F).
Total prey abundance and biomass. High elevation areas with medium vegetation height had the lowest total prey abundance (22 ± 5 individuals per 1,000 m2) and lowest total prey biomass (235 ± 21 g per 1,000 m2). Forested areas with high canopy cover, high ground cover, and tall vegetation had the highest total prey abundance (57 ± 60 individuals per 1,000 m2) and low elevation, medium ground cover, short to medium vegetation height, cropland contained the highest total prey biomass (976 ± 456 g per 1,000 m2). (Fig. 5-6).
Owl presence and pueo occupancy models. Across all sites there were 16 owl detections total during the owl survey period (1 hour prior to sunset to half-hour after sunset), including 11 pueo, 2 barn owl, and 3 unknown owls (barn owl or pueo). The barn owl and unknown owl observations occurred in mid-elevation open areas with low vegetation height and cover. Pueo were observed across a range of vegetation characteristics using both open and forested areas but were most frequently observed at mid to high elevations, in areas with short to medium vegetation height, medium ground cover, and low to medium canopy cover. Pueo responded to audio playback vocalizations at 50 percent of the sites where a pueo was visually detected. At one site a pueo was audibly detected when it responded to audio playback vocalizations at the end of the survey where no pueo were visually detected. (Appendix G).
Pueo pellets. During the field season, we opportunistically found two pueo pellets in a high elevation, shrubby area at Haleakalā National Park. The first pellet contained house mouse bones and the second pellet found nearby contained juvenile rat bones. (Fig. 8).
In this study, we examined both native and non-native potential prey species across three taxonomic groups and a range of vegetation characteristics in relation to pueo occupancy on Maui, Hawai‘i. we expected that the pueo would hunt in areas that were the most accessible (short vegetation and low ground cover) compared to areas with dense vegetation, and that the biomass of prey items most commonly found in pellets (mice and zebra doves) would play a greater role in predicting pueo presence than all potential prey items combined. Pueo were detected across a range of vegetation characteristics but were most often seen in mid to high elevation, using both open and forested areas. The detectability of pueo was weakly positively influenced by vegetation height, suggesting that assumptions regarding a preference of pueo for open vegetation types may be skewed based on the ability of observers to detect pueo in these ecosystems or due to bias from species level information originating largely from select areas throughout the global distribution of short-eared owls. Single-season occupancy models indicated that bird biomass and relative total prey biomass were weakly correlated with the estimated likelihood of pueo occupancy, in other words, that pueo are more likely to occur in areas with lower prey biomass, while elevation was positively correlated with pueo occupancy, indicating they are more likely to occur at mid and upper elevations. Vegetation height and relative insect (Orthoptera) biomass also had a weakly positive relationship with the estimated likelihood of pueo occupancy. The occupancy models do not clearly identify whether vegetation structure (height/cover) or prey biomass are driving distribution but rather, suggest a combination of factors, and potentially others that were not accounted for in this study such as competition or predation risk, influence pueo distribution.
The sites in which pueo were most often observed contained low to moderate biomass of all the prey types for which we surveyed. The models were not driven solely by Microtus spp. biomass, as in the continental United States (Wiggins et al, 2006), but were weakly correlated with insect (Orthoptera) biomass. The two models with the highest AIC weight (combined 65 percent) indicated a weakly negative relationship between bird biomass and relative total prey biomass. Birds have been found in pueo and continental short-eared owl pellets (Holt, 1993; Mostello, 1996; Mostello & Conant, 2018) but require the most energy to hunt and therefore, may not be the optimal prey choice (Toland, 1987). Bird biomass and relative total prey biomass were also highest in low elevation areas where pueo were only detected once. Pueo could be using these areas at a different time of year from when this study was done (May-December) or at a different time of day from the survey window (one hour prior to sunset to one hour after).
Similar to previous studies, rat density was highest in tall vegetation and high canopy cover, and mouse density was highest in short vegetation and low canopy cover (Shiels, 2010; Harper & Bunbury, 2015; Shiels et al., 2017; Tseng et al., 2017). It may be more profitable for pueo to switch between different prey types in Hawai‘i, given that rodent abundance was not as high overall in this study as in other rodent abundance studies globally (Mostello, 1996; Shiels, 2010). Furthermore, an average pueo pellet emitted per day is composed of prey items that are estimated to total 30-40 grams (Clark, 1975). At an average biomass of 11 g per mouse, this indicates that roughly three mice must be consumed daily to meet energetic demands. Thus, prey switching and hunting in diverse habitats may be necessary to meet metabolic and nutritional needs, considering the diverse and available prey biomass across the landscape. For example, the zebra dove (Geopelia striata), a common prey bird found in pueo pellets, is a slow-moving ground feeder with a mean biomass of 55 g, making it an energetically profitable species to capture (Mostello, 1996; Mostello & Conant, 2018). Juvenile birds were also observed in pueo pellets, reinforcing the notion that pueo are opportunistic and predate on energetically profitable prey (Mostello, 1996; Thirgood et al., 2003; Mostello & Conant, 2018). Further, pueo that feed on diverse prey items minimize the potential ingestion of rodenticide, potentially increasing the likelihood of survival and reproductive success. This study did not examine the distribution of reptiles, such as lizards, but future studies should include these potential prey items, as pueo have been observed ingesting these prey items (pers. obs.).
Prey distribution likely varies across vegetation types based on season and rainfall and may drive temporal variation in pueo habitat use. The top predictive models for both the rodents and insects (Orthoptera) contained mean annual rainfall as a covariate (significant for insects), consistent with expectations that prey occurrence is driven by moisture availability (Schmidt et al., 2018). Correspondence with technicians who were rodent trapping for the State of Hawai‘i Vector Control Department reported similar results from trapping taking place concurrent to this study, with few, if any, rodents caught during the summer months at low elevations (Travis Barut, pers. comm.), which was likely due to low rainfall and high temperatures. Pueo, rarely seen on Kaho‘olawe (an island seven miles southwest of Maui), have been noted in abundance during cyclic rodent irruptions in the spring after a rainy winter season (Kaho‘olawe Island Reserve Commission Seabird Restoration Business Plan, 2015). These findings are consistent with other studies that correlate the timing of vegetation growth as stimulation for prey population growth (Banko et al., 2015; Schmidt et al., 2018).
Mid- to high elevation locations had a positive relationship with the estimated likelihood of pueo occupancy and were comprised of a range of vegetation characteristics ranging from short vegetation, low ground cover, and low canopy cover—potentially offering low-effort hunting opportunities—to tall vegetation, high ground cover, and high canopy cover that contained high amounts of prey biomass and may be profitable at times when other areas are scarce in prey. A mosaic landscape of open and forested areas likely provides an energetically profitable combination of prey biomass, accessibility, and perching opportunities (Banko et al., 2015; Shiels et al., 2017). Other characteristics of mid- to high elevation areas on Maui include cooler temperatures, increased connectivity, and little housing or development.
Pueo distribution may be influenced by barn owls or other competitors or predators, as barn owls were observed hunting at sunset and night-time in mid- to lower elevation vegetation types, whereas pueo were observed hunting at daylight until sunset and utilizing higher elevation sites, suggesting that these two species overlap at sunset but primarily occupy two different temporal hunting periods and potentially two different elevational bands on Maui. There have also been reports of barn owls depredating pueo fledglings at the nest (Jake Muise with Maui Nui Venison, pers. comm.). Diet differs significantly among the four potentially competing terrestrial predators in Hawai‘i: the feral cat, small Indian mongoose, barn owl, and pueo (Mostello, 1996; Mostello & Conant, 2018). The two owl species and cats prey primarily on rodents, but the diets of these three species vary by location (Mostello, 1996; Mostello & Conant, 2018). Dietary overlap is highest between the pueo and the barn owl (Mostello, 1996; Mostello & Conant, 2018), though pueo diet is not thought to have changed since the introduction of the barn owl (Mostello, 1996; Mostello & Conant, 2018). On the islands of Hawai‘i (commonly referred to as the Big Island), Maui, Moloka‘i, and O‘ahu, where mongoose are present, both barn owls and pueo consume more insects (Mostello, 1996; Mostello & Conant, 2018). Although there is dietary overlap, competition between the four predators seems unlikely due to variation in distribution, abundance of prey, and the opportunistic nature of foraging by generalists (Mostello, 1996; Mostello & Conant, 2018). However, during times of resource scarcity, competition may increase (Work & Hale, 1996; Mostello & Conant, 2018).
It is important to recognize the value of preserving large expanses of unfragmented habitat used by pueo for hunting and nesting in the upcountry sections of Maui at large historic ranches, natural area reserves, and at Haleakalā National Park. A strategy to promote pueo hunting habitat is to maintain mowing regimes or the rotational grazing of domestic ungulates to retain short vegetation height in locations dominated by invasive grasses. Constructing raptor nests boxes does not benefit pueo and would promote the already prevalent barn owl that potentially competes with pueo. Since pueo nest on the ground and have elaborate courtship displays, it is important to observe and note this behavior that may indicate a nest is nearby. Observing a pueo repeatedly carrying prey (food provisioning) to the same area is an indication that a nest is nearby and a buffer zone of 30-50 meters around the potential nest site is recommended to reduce disturbance. It is also important to keep in mind that pueo fledglings are mobile and walk away from the nest before they can fly, so there is a possibility of encountering young that may seem to be in a random location. It is important to leave them where they are found because a parent is likely aware of the location and feeding them.
Roads are an energetically profitable place for raptors to hunt as there is no protective cover for prey (Bechard, 1982; Preston, 1990), but this has come at a cost as vehicular collisions are a major cause of raptor mortality (Wiggins et al., 2006; Donázar et al., 2016; Miller et al., 2017). An important conservation strategy is to identify sections of roads that are heavily utilized by owls, especially during fledgling season (April-June), and to set lower speeds in those regions with speed bumps and proper signage. Other modifications to reduce collisions with cars and manmade structures would include installation of raptor deterrents on fencing and telephone wire in high-traffic areas frequented by pueo.
Raptors can also be negatively impacted by secondary poisoning from consuming pests that have ingested rodenticide or other pesticides (Work & Hale, 1996; Donázar et al., 2016; Vyas et al., 2017; Mostello & Conant, 2018). After a prey species consumes poison, it may live up to a week and exhibit slower behavior, leading to an energetically profitable capture with potentially negative side effects for the raptor (Vyas et al., 2017). However, predator control of rodents, mongoose, and feral cats by trapping and removal may also benefit the pueo by reducing competition and predation of eggs and young. Utilizing one-way or A-24 traps instead of poison to reduce rodents provides an easy meal to pueo, who have been known to frequent A-24 rat traps to feed on carcasses, and reduces bioaccumulation of pesticides in their system (Franklin, 2013).
The largest cover type on Maui is development (36%) with more land area at lower elevations. Factors influencing nesting success, such as predator abundance, may provide a partial explanation for distribution of pueo in non-developed areas. Potential predators of the ground-nesting pueo, such as cats and mongooses, may occur in higher abundance closer to human dwellings due to increased food resources compared to higher elevation and rural areas (Gaertner et al., 2017). Agencies such as Haleakalā National Park, Maui Forest Bird Recovery Project, and Maui Nui Seabird Recovery Project perform predator control activities to reduce rodent, mongoose, and feral cat populations in higher elevation areas where native and endangered species occur, which may also benefit pueo nesting activity. Targeted predator control efforts for mongoose and feral cats in lower elevation vegetation types, such as in croplands, would improve nesting success for pueo in these areas.
Agriculture producers can play an important role in conserving natural habitats and wildlife through use of beneficial practices (Lindell et al., 2018; McClure et al., 2018). Many farms and ranches in the continental United States play a vital role in wildlife conservation (Maas et al., 2013; Lindell et al., 2018; Heath & Long, 2019; Olimpi et al., 2020). Producers contribute to wildlife conservation partly because it can be profitable for them to do so (“natural” pest control) and partly because they believe it is the ethical thing to do (Maas et al., 2013; Lindell et al., 2018; McClure et al., 2018; Heath & Long, 2019; Olimpi et al., 2020; pers. comm. with ag. producers). Many also value the aesthetics of natural systems and seeing native wildlife thriving on their land (Lindell et al., 2018). Adopting a coexistence model of managing land in Hawai‘i offers a multitude of services—economically, visually, and culturally (Maas et al., 2013; Lindell et al., 2018; McClure et al., 2018; Heath & Long, 2019; Olimpi et al., 2020).
For pueo in cropland and grassland areas, an agroecosystem management approach could be highly beneficial. In cropland, prey species were localized near water sources and active cultivation; by recognizing water resource locations, efforts can be focused on reducing potential pests in that area. The presence of an owl reduces the activity of pest species; thus, maintaining perch sites such as posts, trees, and hedgerows, or uncultivated areas for a pueo to nest, will aid in promoting their use across vegetation types. With the conversion of fallow sugarcane to active cropland, pests and their predators are likely to increase, which may provide an opportunity for agriculture operators to further benefit from pueo (Koopman & Pitt, 2007). Pueo consume a large number of non-native species that can become rampant pests. By minimizing the use of poison and encouraging habitat for owls, while also keeping larger invasive species (mongoose, cats) under control, land managers allow pueo to continue to provide a service to native ecological systems by reducing populations of invasive birds, rodents, and insects (Orthoptera).
Outreach to the general public and agricultural producers regarding conservation implications is necessary to improve the understanding of pueo identification, life history, and behavior, in addition to the many benefits of having pueo present across the landscape. Outreach can also increase engagement in best management practices and modifications to provide ideal habitat for pueo.
3. Determine seasonal habitat use by Pueo.
A total of 85 trapping sessions were conducted over a 7-month period from February 2019 through February 2020 for a total of approximately 1300 person hours. During this period, we captured and banded a total of 5 pueo (2 adult females, 1 subadult female, 2 adult males) on O‘ahu, and VHF transmitters were attached to all owls except for one adult female (Table 3.3). VHF tagged individuals were then tracked and resighted for an average of 53 days until the tags failed, or the individual died (Table 3.3). We recorded an average of 26 resight locations per individual; however, due to a harness failure (adult male) and one mortality (subadult female), two individuals were tracked over a relatively shorter period. An additional 2 fledglings of unknown sex were hand-captured at nests incidentally discovered in March of 2020. One fledgling was banded and released and the second was released without bands because it was too young to be banded.
An adult male tagged with a VHF transmitter on 11th of February 2019 was able to destroy the harness and the transmitter was recovered 7 days later. Owls in general are well known to be hard on transmitters due to their mobile necks and hooked bills allowing them to access tags and harnesses in ways that other species cannot. The adult male has subsequently been resighted in the same area exhibiting normal behavior. A subadult female, tagged on 29th of April 2019, was found dead 23 days later on the 22nd of May 2019 in an open field in close proximity to a communications tower (21°25’12.7″N 158°08’49.2″W). The individual was observed alive the previous night putting time of death late May 21st or early May 22nd, 2019. The body was collected and transferred to Dr. Thierry Work at the United States Geological Service National Wildlife Health Center Honolulu Field Station for necropsy. Necropsy results indicate trauma as the likely cause of death. Apparent cause of death coupled with the location of the body near the communication towers indicate that collision with the tower or guy wires likely resulted in death. Some feather loss was noted under both wings near the harness strap.
The remaining two tagged pueo, an adult male and adult female, were tracked and resighted for an average of 90 days, until the internal batteries of the VHF transmitters became too weak to emit a signal (Table 3.3). I recorded an average of 46 resight locations for these two individuals. Resight locations were recorded between the hours of 0500 Hawai‘i Standard Time (HST) and 2300 HST. At the time of resight, I observed individuals using land-cover classes including developed (n = 25), grassland (n = 41), kiawe woodland (n = 22), and urban (n = 4; Table 3.4). Only one individual was observed using urban areas in the town of Māʻili about 1.5 km from its roost site. Kernel density home range estimates were generated for these two individuals with an average 50% core area of 1.12 km2 and an average 95% full area of 5.57 km2. Land-cover composition of home ranges are reported in Table 3.4. Single occasion forays to new locations up to 4.5 km away from the full home ranges were recorded four times and in all cases the individual returned to their home range within a day.
Behaviors observed at the time of resight included roosting (n = 34), hunting (n = 32), agonistic (n = 2), calling (n = 2), flying (n = 1), and unknown (n = 21). Individuals were observed hunting both aerially (n = 4) and from perches (n = 28) including the ground (n = 19), trees (n = 5), and streetlights (n = 4). We observed individuals roosting both perched in trees (n = 29) and on the ground (n = 5). Few resights were recorded in urban areas (n = 4) and the urban land-cover class was excluded from analysis, leaving developed, grassland, and kiawe woodland land-cover classes. Using a Pearson’s chi-square test we found land-cover class use was significantly associated with behavior for the two individuals with >40 resight locations pooled (P < 0.01; χ2 = 43.7). A post-hoc analysis indicated that hunting behaviors were more often observed in grassland (83%) than developed (14%) or kiawe woodland (3%) land-cover classes. Roosting individuals were more often observed in developed areas (53%) in comparison to kiawe woodland (44%) or grassland (3%; Table 3.5). The unknown behavior category showed no significant association with land-cover class, indicating that the frequency of unknown observations was not related to land-cover (Table 3.5).
We recorded resights most often during diurnal periods (n = 46) followed by crepuscular (n = 27), and nocturnal (n = 19) time periods. Using a Fisher’s exact test, we found that time period was significantly associated with activity for the two individuals with >40 resight locations pooled (P < 0.01). A post-hoc analysis indicated that the proportion of resights of active individuals was highest during nocturnal (70%) and crepuscular (63%) periods and lowest during diurnal periods (15%; Table 3.6). The unknown behavior category showed no significant association with time period, indicating that the frequency of unknown observations was not related to time of day (Table 3.6).
Short-eared owls are typically considered open habitat specialists, utilizing grasslands, shrublands, and agricultural areas almost exclusively, but our resight data from VHF tagged owls in Hawai‘i allowed us to document consistent use of forested and wooded developed areas in addition to grasslands. Although we are unable to make broad population inference due to the small sample size of this study, these findings hint at a more diverse range of habitat used by pueo than their continental counterparts.
The two tagged pueo occupied adjacent territories throughout the duration of the study and home range estimates built for the two individuals showed considerable overlap at the full and core use areas (95% and 50% contours, respectively). The two individuals were observed using the same areas, including even the same specific trees, for roosting consistently throughout the study period. At one point the two pueo were observed roosting in the same tree as a barn owl. Home range overlap of the two was also evident in grassland areas used for hunting nearby, however, the owls defended hunting patches and when one individual encroached on the hunting territory of the other a fight would ensue. On occasion if an individual began hunting before the other, they would attempt to hunt in adjacent hunting territories only to be chased out when the owner left their roost.
Hunting and roosting behaviors were significantly associated with land-cover class use for these two individuals, with hunting occurring mostly during crepuscular and nocturnal periods in grassland, and diurnal roosting in kiawe woodland and developed areas. Strikingly, one individual (male) was found occasionally leaving the area, flying approximately 1.5 km over the town of Māʻili, to hunt in a busy urban beach park. During one observation we resighted the pueo hunting from a perch in a palm tree in the beach park.
Activity levels of the two tagged pueo were correlated with the period of day. Active behaviors, including hunting, agonistic, callings, and flying, peaked during crepuscular and nocturnal periods while roosting peaked during diurnal periods. Neither of the pueo were observed roosting during nocturnal periods, and were only infrequently observed active during diurnal periods.
Short-eared owls in general are highly volant and capable of inter-island movements. In the Galápagos researchers found evidence for interisland movements by the endemic Galápagos short-eared owl (Asio flammeus galapagoensis) over distances similar to those found in Hawai‘i (Schulwitz et al. 2018). Pueo have also been observed breeding on the island of Kaho‘olawe during annual rodent irruptions and conspicuously absent from the island when rodent abundance is low, suggesting regular movement to and from the island (J. Bruch, Kaho‘olawe Island Reserve Commission, personal communication). In our study, however, we consistently resighted VID tagged pueo in the same territories for two consecutive years after banding. This hints that at least some of the population may be largely resident in Hawai‘i, and further research is needed to explore movement patterns over longer periods of time and at more sites with the use of solar powered GPS devices.
4. Develop Habitat Conservation Recommendations for Producers.
Habitat conservation recommendations are detailed above in the context of the other studies, but can be summarized as follows:
- Pueo are more active during the crepuscular and nocturnal periods than during diurnal periods, and thus may be more driven by prey availability during these times of day. Cattle are often crepuscular grazers as well, so awareness of preferred vegetation overlap areas is important. Potential management actions would be to rotate cattle away from known pueo habitat during nesting season, or to temporarily fence known nest areas during that period.
- Pueo are more likely to be observed in open habitats (agricultural lands, grasslands), but this may be due to vegetation inhibiting detectability in taller vegetation. However, hunting of prey should be easier in open habitats; thus agricultural lands with high prey abundance are likely to provide suitable habitat for pueo. Adjacent perches in trees, shrubs, and particularly fenceposts and utility poles, is helpful. Maintenance of hedgerows and non-grassland edge habitat is likely helpful for perch provision.
- Pueo were observed to be fairly cryptic, particularly when nesting, and would spend substantial time resting or hunting from the ground. Further, chicks may move as much as 100 m away from nest sites prior to fledging, and will remain on the ground when approached, risking crushing by stepping on them or machinery. As such, operators of heavy machinery should be trained to immediately stop machinery if a pueo is flushed, and search the area for nests and chicks to avoid destroying them. Again, ranchers practicing rotational grazing are recommended to fence cattle away from nests during nesting season.
- Since pueo have high site fidelity (establish home range and remain in the area year-round), impacts from rodenticide use or other on-site threats are more likely to impact them year-round, compared to continental Short-eared Owls that are highly vagrant and seasonal. Consider ladders in water troughs to minimize drownings, minimize or cease use of rodenticides, and/or use bait/traps that do not allow rodents to return to the landscape after ingesting bait, and/or incorporate non-bait stun techniques such as GoodNature A24 repeater traps.
- Consider mowing and rotational grazing regimens that minimize landscape disturbance in known owl areas, as pueo appear to prefer mid-height grasses for nesting, and mowing disturbs and/or destroys nests.
5. Increase producer and public awareness of the role of Pueo in agriculture.
Best management practices were presented to producers during several in-person stakeholder meetings in 2019 and via Zoom in 2020 and 2021, including mowing length recommendations, current understanding of preferred nest-site characteristics, and estimated prey items removed by nesting and non-nesting owls. Workshop and presentation locations included various sites on both O‘ahu and Maui (detailed earlier), as well as the Pueo Project website, booths at agricultural and community events, and the Symphony of the Hawaiian Birds presentations across the Hawaiian Islands. Total number of direct producer interactions were at least 10-15 unique conversations, and presentations reached at least 100 producers . Total public interactions across all of these events was at least 1,600.
Educational & Outreach Activities
Educational site visits with farmers and community:
We visited stakeholders in October 2018. The initial stakeholders included Alexander & Baldwin, Haleakala Ranch, and East Maui Watershed Project. We visited each stakeholder at their property and took a tour of the area. We discussed the needs and expectations of the partnership, pest issues, and the survey plan. We provided materials for employees to record any observations to understand what areas of the property pueo are seen. We re-visited these site in February 2019 and 2020 to follow up with stakeholders to answer any questions, discuss areas of their property accessed for surveys, and provide a field ID guide for stakeholders and their employees to use for field-identification purposes.
We also communicated with land managers in 2018-2019 across the Hawaiian Islands to discuss threats to pueo and historic mortality data: Kauai Seabird Recovery Project, Maui Nui Seabird Recovery Project, Hawai’i Wildlife Center, and the Airports.
We communicated with interested volunteers and farmers through the Pueo Project website (www.pueoproject.com). We received useful reports on pueo activity across the Hawaiian archipelago. Our lab made a connection with Brynn Foster, owner of Voyaging Foods Farm on O’ahu, who observed pueo and was interested in the Pueo Team visiting her land to recommend habitat preferences of pueo. We also spoke with Bronwyn from Molokai’s farm sanctuary, Hui Ho’olana; she wanted to promote pueo habitat on the sanctuary and wanted the Pueo Team to visit and make further recommendations.
The Pueo Team (Dr. Price, Dr. Cotin, Prof. Steensma, Chad Wilhite, and Laura Luther) met with Bayer (formerly Monsanto) to discuss a partnership to access their lands to monitor pueo and provide feedback on habitat preferences and the feasibility of pueo as a biocontrol. We visited Bayer’s main property on O’ahu on Friday, October 19, 2018 for a tour of the landscape to identify potential pueo habitat and vantage points to survey from. We also left material with Bayer employees to report any observations. In 2020 we met with Bayer employees to provide an informational training about pueo observation and reporting during a ‘Lunch and Learn’ session.
In 2019 we met with Mahi Pono, growing a variety of crops on former sugar cane land on Maui, to discuss pueo habitat and access points there. We also met with managers at Haleakala Ranch and Ulupalakua Ranch, all on Maui, to discuss habitat use and survey access points for the upcoming field season.
In October of 2019 our team conducted survey protocol training on Oahu, Maui, Kauai, and Hawaii islands to train forestry and wildlife personnel in pueo survey techniques and identification.
We presented to the Oahu Farmer’s Association in the fall of 2019 to talk about pueo as biocontrol on producer lands. In February 2020 we presented results of the Maui surveys to managers at Haleakala Ranch. We also presented results of the Maui surveys to a group of producers and Haleakala National Park employees, at Ulupalakua Ranch. We also presented results of the Maui surveys to a meeting of the Haleakala Chapter of the Hawaii Farmers Union. We also presented to leaders of the Maui Conservation District and the Maui Farm Bureau. In all presentations we talked about pueo as biocontrol on producer lands, as well as valuable native species and cultural aumakua. All of these presentations included discussion time and management ideas for protecting pueo, and for enhancing farm and ranch profits through provision of pueo habitat and resulting decreases in insect, bird, and rodent pests.
In May and November of 2018, and October of 2019, the Pueo Team had an educational booth, along with other wildlife organizations at the Hawaiian Bird Symphony Event (~6000 attendees). We had a museum specimen of pueo and talons to show children and provided material regarding pueo identification and current research efforts. We had an educational booth at the Bishop Museum’s family science day in March of 2019, in which approximately 500 children and family members participated.
In November of 2018 our team was interviewed for an article on pueo in Maui Magazine, reaching thousands of readers on Maui. https://www.mauimagazine.net/hawaiian-owl/
In December 2019 our project was featured in the Pacific Birds newsletter, reaching hundreds of bird enthusiasts. https://pacificbirds.org/2019/11/finding-the-pueo/
Changes in knowledge, awareness
(1) times of day in which pueo may be observed
(2) potential vegetation types where pueo nests may be found
(3) potential diet of pueo
Farmers were largely very willing to adjust practices to promote pueo on their lands. Most farmers already knew how to identify pueo and valued them culturally, personally, and ecologically. Some needed information about distinguishing between pueo and introduced barn owls. All were interested to learn the degree of pest control provided by pueo against rodents, insects, and pest birds. Areas of potential growth are in decreasing the use of pesticides and rodenticide, which are likely to impact pueo through secondary poisoning, based on a recent USDA report that examined presence of these chemicals in owls in Hawaii. Farmers are concerned about balancing a desire for owls on their property with the need to control pests at large scales. An area of great interest that would benefit from a cost-share program, would be availability of A24 traps for killing rodents through stunning, rather than rodenticide. Farmers were also interested in the potential for drowning of owls in water troughs, a cause of mortality which has been observed on the mainland but was not directly reported by farmers in this study. Farmers were open to adding simple ladders to the inside of troughs to decrease the likelihood of drowning if owls attempt to drink from, or land in, troughs. Most ranchers were already practicing careful rotational grazing practices and conservation of edge habitat, but were willing to add protection of owl nesting season to their grazing, fencing, and mowing regimes. Finally, invasive mammals such as mongoose, cats, and dogs are likely causes of nest failure. Responses varied among farmers in regards to managing these threats to nesting success and adult survival, but some producers — especially those closer to residential areas — were willing to participate in feral cat trapping programs.
Overall, the economic benefits of healthy pueo populations in agricultural areas were demonstrated to producers and the public, through prey data collected during this project. Removal of insects, pest birds, and rodents is directly related to increased production across the agricultural landscape. Environmental benefits for farmers will also ensue, as their ability and desire to retain hedgerows, wooded edges for owl perch sites, and medium height grass areas for owl nests, will also improve conservation values for their land and even prevent soil erosion when combined with riparian protection and protection from overgrazing. The social benefits for farmers include not only increased viability of their own economic situation, but also pride in their conservation measures and joy in seeing these native raptors on their land. Their ability to share in promoting esthetic and conservation values with the community for a culturally valued species, improves their leadership and social standing in the community as well. This project has been a “win-win-win” for conservation science, agricultural sustainability, and communities in the Hawaiian Islands.
- Annual Report to the Hawai‘i Division of Forestry and Wildlife
- Laura Luther M.S. Thesis: Factors influencing the distribution of the Hawaiian short-eared owl (pueo) on the island of Maui, Hawai‘i
- Chad Wilhite M.S. Thesis: Population dynamics and habitat use of the pueo (Hawaiian short-eared owl; Asio flammeus sandwichensis) (not yet available, defended thesis May 2021) pdf
- Poster presentation at Raptor Research Conference pdf
Multiple ranchers described seeing the same pueo on a post or hunting on-the-wing most days while driving into the field in the morning or evening.
A manager on a large farm that is largely fallow currently described practices on their active lands that minimize heavy equipment impacts to nesting pueo.
Multiple producers expressed feeling kinship feelings regarding pueo, and valued observing them flying or perching on their property. Many considered these to be “their” owls, or even described them as aumakua (ancestral guardians).
Mae Nakahata, long-time sugarcane geneticist with Alexander & Baldwin who had first contacted a member of our team primarily to request support for control of problem birds in agriculture, shared a beautiful photo taken not long after one of our discussions on improving pueo habitat. Her expression of pride and value of pueo on the landscape was gratifying.
Assessment of research and educational methods. Survey methods developed for this project were sufficient to meet goals, but substantial time and effort was involved, and limited the total number of site visits that we could conduct on each island. This will be an ongoing challenge in studying pueo, as they are broadly distributed across multiple habitat types, but in low abundance. For similar reasons, capture rates for pueo were very low, limiting the number of birds we could track for the study. A subsequent funding source has allowed us to use GPS-VHF devices, minimizing the time spent tracking birds, and we have had increased trapping success through the use of new traps (dhou gaza) and lures (decoys of pueo seem to work better than decoys of prey or competitor/predators), at least for breeding pueo. Again due to low abundance, discovery of pellets was minimal during the study period. Pueo pellets and barn owl pellets can look similar, especially if they are older, and thus unless pellets are discovered in a perch that is regularly and solely used by pueo, one cannot be sure which species they are from. However, the survey methods to determine distribution of potential prey, and pueo occupancy in response, were very useful. We do note that due to predator-prey cycles, pueo are likely to occupy locations with higher abundance of prey items. However, because they can prey switch, they may maintain occupancy at particular sites by switching among rodents, birds, and invertebrates, thus providing continuous biocontrol of various species on agricultural lands.
One of the greatest benefits of this project was the leveraging of funding from several sources to accomplish work that could not have occurred with any one of the funding sources. Another benefit was the development of good working relationships with producers across the islands. Trust between conservation scientists and agricultural producers is an important outcome of this project.
Future endeavors. As many of these potential pueo prey items are also pest species negatively impacting farms, we hope this will be useful information to farmers in communicating potential for pest control by pueo. Moving forward, farmers expressed quite a bit of concern about invasive ungulate impacts on their lands. They really enjoy seeing pueo, and recognize their value as biocontrol (in part thanks to our project), but are experiencing major economic impacts to crop production and cattle production due to damage from invasive pigs, invasive deer, and invasive goats. Thus, we have been working over the last year to develop collaborations and proposals that will fund research and education on ungulate impacts, potential removal solutions, and potential collaborations across agencies, landowners, and hunters. We believe that such ongoing efforts will also aid pueo survival, as invasive pigs eat owl nestlings and destroy nest sites, and other ungulates trample pueo nests and young, and also overgraze, destroying pueo habitat.