Monitoring Population Size of Endangered Hawaiian Duck and Prevalence of Mallard/Hawaiian Duck Hybrids on Hanalei and Hulē‘Ia Nwrs Kaua‘I, Hawai‘I Final Report

Monitoring population size of endangered Hawaiian duck and prevalence of Mallard/Hawaiian duck hybrids on Hanalei and Hulē‘ia NWRs Kaua‘i, Hawai‘i Final Report

Koloa maoli female (Photo M. Couffer/USFWS)

Submitted 1 September 2018: Inventory and Monitoring Program Initiative, Region 1, USFWS

Submitted by:

Dr. Bruce D. Dugger Christopher P. Malachowski Department of Fisheries and Wildlife Department of Fisheries and Wildlife Oregon State University Oregon State University 104 Nash Hall 104 Nash Hall Corvallis, OR 97331 Corvallis, OR 97331 Phone: 541-737-2465 Phone: 215-850-1978 E-mail: [email protected] E-mail: [email protected]

Kimberly J. Uyehara Kaua‘i National Wildlife Refuge Complex U.S. Fish and Wildlife Service PO Box 1128 Kīlauea, HI 96754 Phone: 808-635-9309 E-mail: [email protected] ‘‘The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.’’ TABLE OF CONTENTS

1. EXECUTIVE SUMMARY………………………………………... 1

2. INTRODUCTION………………………………………………….. 3

a. Objectives………………………………………………….. 4

3. STUDY AREA……………………………………………………… 5

4. METHODS………………………………………………………….. 5

5. RESULTS…………………………………………………………… 14

6. DISCUSSION……………………………………………………….. 19

7. ACKNOWLEDGMENTS…………………………………………… 26

8. LITERATURE CITED………………………………………………. 27

* All photos credited C. Malachowski unless otherwise noted

ii EXECUTIVE SUMMARY

The Hawaiian duck (hereafter Koloa) was listed as an endangered species in 1967. Key threats to species recovery are the lack of basic demographic information, like population size that can be used to inform recovery decisions, and genetic introgression with feral Mallards. Here we report on the results of a joint research project and conservation action to achieve the following objectives: 1) investigate the value of using banded birds and a mark-resight analysis approach to estimate the population size of Koloa using Hanalei and Hulē‘ia NWRs; 2) Determine the prevalence of Mallard/Koloa hybrids on Hanalei and Hulē‘ia NWRs; 3) Collect data to help refine the Mallard/Koloa hybrid key to improve implementation of hybrid removal activities at both Kaua‘i refuges and James Campbell NWR on O‘ahu; 4) Conduct field surveys at Hanalei and Hulē‘ia to survey for extent and seasonality of intra-island movements; and 5) Develop operational survey to annually detect and remove hybrids from Hanalei and Hulē‘ia NWRs; explore application of these methods to James Campbell NWR on O‘ahu. Between December 2010 and December 2012 we captured and banded 835 unique Koloa, most at Hanalei NWR (97.0%); 78.4% of captured birds were male. The male-bias in the sex ratio was higher than previously reported for any continental dabbling duck, was consistent with other measures of the sex ratio at Hanalei NWR, and is something that should be investigated. We detected no band loss or excessive wear looking at 918 recaptures through November 2015. Persistent banding increased the average number of banded birds detected on surveys from 6.2 in 2011 to 34.5 in 2013. Of those birds that were determined to be banded, we were able to read the alpha numeric code for just over 39% of bands in 2011 and 52% in 2013; however, for most birds it was difficult to read bands during visual surveys. Nine captured males (1.1% of our capture sample) scored as hybrids using a plumage and morphometric key and were subsequently humanely euthanized. Subsequent genetic testing assigned 8 of those 9 birds a probability of 95% being Koloa. We convincingly demonstrated that a large number of Koloa can be safely captured using a traditional trapping technique developed for waterfowl, the baited swim-in trap. However, the considerable survey effort needed to detect and read alpha-numeric color bands in the field indicated that a band-resighting will not work for the surveyor who needs to efficiently cover many sites in a single day or for busy refuge biologists who do not have the time available for band searching with considerable intensity. Thus, we recommend future efforts to estimate population size or survival using banded birds adopt a capture-recapture methodology. The

1 occurrence of Mallard-Koloa hybrids at Hanalei NWR is very low, a conclusion supported by subsequent genetic testing of a large sample of our captured birds. Previous estimates for the prevalence of hybrids on the island have been based on observations of “hybrid-looking” birds seen during periodic surveys. Our work on the plumages and molts in conjunction with collaborators at the University of California Davis has revealed significantly more variation in male plumages than previously described for Koloa. This variation is known to occur in birds that are genetically confirmed to be Koloa; thus, any estimate of hybrid prevalence based solely on visual sightings will likely be biased high for males. Continued application of the existing plumage/morphometric key would result in false positives for 1% of birds (i.e., they score as hybrids when genetically they score as Koloa). Given examples of variation in culmen measurements among measurers and among attempts for a single measurer on a single bird, a more risk adverse method (e.g., minimize false positives) of using the morphometric key would involve building in measurement error to the decision making threshold criteria values. Despite banding more than 1,000 Koloa on Kaua‘i since 2010, and extensive survey activities on Oʻahu during summer 2015, no birds banded on Kaua‘i have been observed on O‘ahu. We conclude the report by describing a protocol that could be used to implement a capture-recapture program at Hanalei NWR to estimate Koloa population size and possibly survival.

INTRODUCTION

The Hawaiian duck or Koloa maoli (Anas wyvilliana, hereafter Koloa) is endemic to Hawai‘i and one of three extant waterfowl species occurring on the Hawaiian Islands (Olson and James 1982, Engilis et al. 2002). Relatively common during the early 1900s, the Koloa has since experienced a significant statewide population decline because of factors that include wetland loss, over harvest by hunters, and introduced mammalian predators. More recently, the genetic integrity of the species has been threatened by hybridization with feral Mallards (Engilis et al. 2002). Consequently, Koloa was listed as a Federally-endangered species in 1967 and has the highest recovery priority for the four listed wetland birds that occur on the main Hawaiian Islands (USFWS 2011). Kaua‘i is the only island that likely supports a population of non-hybridized Koloa (Wells et al. 2016), and Hanalei and Hulē‘ia National Wildlife Refuges (NWRs) are the number one and two most important sites for Koloa in the state.

There is currently no credible estimate of Koloa population size at any spatial scale (Engilis et al. 2002). Although state biannual counts may provide an index for wetlands, counts do not include streams where many Koloa are believed to reside on Kaua‘i (Schwartz and Schwartz 1953, Swedberg 1967). The state biannual count peaked at 524 for the all-island survey and 459 for Kaua‘i in January 2004 (Hawai‘i DLNR unpublished data 1986-2006). However, during banding activities from December 2010 to March 2011 on Hanalei NWR we banded 291 individual Koloa and resighting activities indicated that greater than 75% of birds on the refuge were not banded (C. Malachowski unpubl. data), which indicated the count data were likely biased low. Additionally, limited radio telemetry data from Hanalei NWR (C. Malachowski unpubl. data) indicate there are common movements of Koloa between Hanalei NWR and surrounding areas indicating that mid-day counts on the refuge likely are not adequately characterizing the role that Hanalei NWR plays in meeting the daily habitat needs for Koloa in the Hanalei region. These facts point to a need to investigate other methods for estimating Koloa population size.

Beginning in December of 2010, we initiated a joint conservation action and research project to begin identifying and removing hybrids from Hanalei and Hulē‘ia and investigate basic aspects

of Koloa ecology that could aid in species recovery including movements, survival, nest success and habitat use. Our initial success with trapping and banding Koloa (mentioned above) suggested that capture-recapture or mark-resighting methodologies might provide a means of estimating annual survival and regional population size as well as helping understand movements of Koloa around Kaua‘i and between islands. However, capturing and marking a sample of birds is only one step to make such calculations possible. Additional fieldwork to determine band retention rate, encounter rate with banded birds, and determine our ability to read bands in the field was also needed. Using funds from USFWS (Region 1 Refuges, Migratory Birds and Inventory and Monitoring and Oregon State University and in-kind contributions from the Kaua‘i NWRC) we continued our conservation effort to identify and remove hybrids and expanded our effort to survey for banded birds on Kaua‘i. Our objectives would help meet multiple monitoring and research objectives listed in the draft Hanalei and Hulē‘ia NWR Comprehensive Conservation Plan (CCP) and Hawaiian Waterbird Recovery Plan. The protocol for our survey also allowed us to move forward with key conservation and research activities that supported implementation and assessment of refuge management plans and inform refuge staff about the value of habitat management programs on both refuges.

Objectives: 1. Investigate the value of using banded birds and a mark-resight analysis approach to estimate the population size of Koloa using Hanalei and Hulē‘ia NWRs. 2. Determine the prevalence of Mallard/Koloa hybrids on Hanalei and Hulē‘ia NWRs. 3. Collect data to help refine the Mallard/Koloa hybrid key to improve implementation of hybrid removal activities at both Kaua‘i refuges and James Campbell NWR on O‘ahu. 4. Conduct field surveys at Hanalei and Hulē‘ia to survey for extent and seasonality of intra- island movements. 5. Develop operational survey to annually detect and remove hybrids from Hanalei and Hulē‘ia NWRs; explore application of these methods to James Campbell NWR on O‘ahu.

The reporting period for this grant was March 2011 to July 2012; however, where possible we report on results related to achieving each stated objective through December 2016 to provide

USFWS a more complete overview of progress related to our stated objectives. We organize our narrative around the five main objectives of the project.

STUDY AREA

The study was centered at Hanalei NWR on the north shore of Kaua‘i County, Hawaiʻi (22.205° N, 159.475° W). Hanalei NWR was established in 1972 under the Endangered Species Preservation Act to aid in the recovery of endangered Hawaiian waterbirds through the protection and management of core habitat. The refuge supports breeding populations of five endangered species including the Koloa, Hawaiian coot (Fulica alai), Hawaiian goose (Branta sandvicensis), Hawaiian common gallinule (Gallinula galeata sandvicensis), and Hawaiian stilt (Himantopus mexicanus knudseni). Hanalei NWR provided easy accessibility, limited human- associated interference, and the highest probability of successfully capturing Koloa (Uyehara 2007).

The Hanalei River headwaters form on Mount Wai‘ale‘ale (1,569 m elevation) and flow 25.2 km north to Hanalei Bay. The lower 5.6 km section of the Hanalei River flows through Hanalei NWR where water is diverted to taro lo‘i and managed wetlands. Precipitation at Hanalei NWR varies between a dry season (10.9 – 16.4 cm/mo from May to October) and a wet season (17.3 – 23.1 cm/mo from November to April); mean annual rainfall at Hanalei NWR is 208.8 cm/yr (NCDC 2012). Temperatures at Hanalei NWR fluctuate very little throughout the year. The mean high temperature is 25.3° C in February and 29.4° C in August; the mean low February and August temperatures are 17.2° and 20.7° C, respectively (NCDC 2012).

METHODS

Achieving our objectives required that we capture and mark a sample of Koloa. We used baited swim-in traps (Hunt and Dahlka 1953) to capture Koloa at Hanalei and Hulē‘ia NWRs between December 2010 and December 2012. We customized a conventional swim-in trap design (Hunt and Dahlka 1953) to minimize chance of harm to Koloa and non-target species. Traps were constructed using welded wire mesh (2.5 cm x 5 cm; Dubois 1975). The circumference of traps

varied from 8 to 12 m, and the height of traps ranged from 1.2 to 1.5 m. A panel of 0.6 cm hardware cloth (i.e., wire mesh) was wrapped around the bottom 0.50 m of all traps and attached with plastic zip ties. The hardware cloth deterred fish, snails, and frogs and reduced the risk that waterbirds would entangle their bill, head, or legs. All edges and points on the traps were filed to create a smooth surface that limits abrasions on captured birds. To further minimize possible abrasions, all zip tie clasps were positioned on the outside of traps. We packed the inside lower edge of the trap wall with mud, to prevent birds from getting pinned between the trap and mud when they dove, and we created a small island using mud or logs that provided a place for birds to sit out of the water. Bird netting was then placed on top of the traps and sealed down with zip ties placed 2 to 5 cm apart.

We selected trap sites based on bird density, level of human disturbance, accessibility, vegetation cover, and water depth. We avoided areas with comparatively higher levels of human disturbance and Hawaiian gallinule abundance. We placed traps at sites in water that ranged from 15 to 50 cm deep in areas with no emergent vegetation, which helped minimize gaps between the trap and the wetland bottom, further reducing the risk of injuries. If vegetation was present, small pockets were cleared before traps were placed in wetlands. Recently flooded wetland management units appeared to be ideal trap locations.

After a site had been selected for a trap, it took 7 to 10 days of preparation work before trapping began. We baited capture locations with wheat or cracked corn several days prior to deploying traps. Swim-in traps were then installed, unset (or inactive) with no top netting, and baited for an additional four days prior to capture efforts to acclimate birds to the trap (Fig. 1). We initially checked traps twice per day, once during mid-morning and once just prior to sunset. However, after a Hawaiian gallinule mortality event, we modified the trap checking protocol to include two checks per day. The first check occurred during early morning, within one hour of sunrise, when birds were removed from the trap and banded. The trap was then closed until late afternoon (two hours before sunset), when we reopened the trap (Fig. 2). The last visit occurred after sunset to assure that no non-target organisms had been captured during the late afternoon.

We carefully removed trapped ducks by entering the trap from an access door built into the side of the trap (Fig. 3) and used either our hands or a long-handled dip net to capture birds. The dip net mesh size was approximately 1.3 cm to prevent entanglement of legs, bills, and wings of ducks in the netting. We immediately released all non-target endangered birds. All Koloa captured were placed in poultry crates that were covered with a thin, dark cloth to provide security and reduce stress (Fig. 4). Crates were placed in a sheltered location to protect birds from the sun or rain as they waited to be processed.

Figure 1. Initial set of bait trap with wide access and no roof netting that permitted free access of ducks to the bait, which was located in the center of the trap.

Figure 2. Photos of completed and set bait trap.

Figure 3. Access door built into the side of the trap.

Figure 4. Removing trapped Koloa from the trap (Photo M. Couffer/USFWS).

1. Investigate the value of using banded birds and a mark-resight analysis approach to estimate the population size of Koloa using Hanalei and Hulē‘ia NWRs.

Each bird classified as a Koloa was aged and sexed using plumage characteristics (Engilis et al., unpubl. data) and cloacal examination, and banded with a size 6 or 7a standard USGS metal leg band (Gustafson et al. 1997). Males were banded with USGS bands on the left leg and females on the right. When available, Koloa were also banded with a colored leg band with a unique combination of alpha-numeric characters on the other leg. The color bands applied to birds at Hanalei were black writing on yellow background for birds captured from December 2010 to June 2012 and black writing on white background for those captured in December 2012 (Fig. 5). The change in color scheme was required due to a lack of yellow band material by our supplier. Birds captured at Hulē‘ia wore bands with a black background and white writing. Banding data was provided to the Bird Banding Lab and Fish and Wildlife Biologist Aaron Nadig of the Pacific Islands Fish and Wildlife Office (PIFWO) who maintains the master waterbird list for the Hawaiian Islands.

We used our sample of double marked birds (metal + color band) to estimate band retention for recaptured birds and look for bird movements away from Hanalei NWR. We conducted seven monthly surveys around Kaua‘i (excluding Hanalei NWR) from February to August 2011. During each survey, we visited 20 to 24 sites which included a variety of wetland habitat types (e.g., managed wetlands, taro lo‘i, streams and rivers, reservoirs, pastures, artificial ponds; Fig. 6). We also conducted nine band resight surveys at Hanalei NWR between February and August 2011 and four surveys between January and April 2013. Surveys were conducted from car and on foot with birds being viewed using binoculars and spotting scopes. For each bird seen we recorded if we could determine its band status (banded v. not banded) and the code on the color band if readable.

Figure 5. Picture of colored leg bands applied to birds at Hanalei NWR (top) during December 2010-June 2012 and Hulē‘ia NWR (bottom) from December 2010-December 2012.

Figure 6. Locations of sites included in island wide surveys to look for banded Koloa on Kaua‘i.

2. Determine the prevalence of Mallard/Koloa hybrids on Hanalei and Hulē‘ia NWRs.

We tested a morphometric key developed by Engilis and Eadie (Oral presentation to USFWS and Department of Land and Natural Resources Officials, 2009) to classify each bird captured as either a Koloa or Koloa-Mallard hybrid. All birds determined to be Mallards or Koloa-Mallard hybrids were humanely euthanized following guidelines stipulated in The Use of Wild Birds in Research (Fair et al. 2010), a federal endangered species permit issued to the Kaua‘i NWRC (KNWR-7), a federal migratory bird depredation permit issued to the Kaua‘i NWRC (120267), and approved by the Institutional Animal Care and Use Committee at OSU. When a bird’s identity was uncertain, the bird was treated as a Koloa.

We estimated the percent of Koloa-Mallard hybrids in the population at Hanalei by dividing the number of birds classified as hybrids by the total number of individual ducks (Koloa, Mallards, and hybrids) captured.

3. Collect data to help refine the Mallard/Koloa hybrid key to improve implementation of hybrid removal activities at both Kaua‘i refuges and James Campbell NWR on O‘ahu.

There are some Koloa-Mallard hybrids, particularly Koloa morphotypes, that cannot be accurately determined based solely on plumage and morphometrics (A. Engilis, Univ. of California, pers. comm., 2010). More samples were needed to validate the male hybrid key and increase confidence level in the female hybrid key. Consequently, we collected 10-17 measures of structure and plumage from each Koloa and hybrid capture. We also took a series photographs of various plumage regions for each bird. Finally, we collected blood samples in two hematocrit tubes (<1% of blood volume) from the tibiotarsal vein from all hybrids that were euthanized and from a sample of birds determined to be pure Koloa. Blood and tissue samples were sent to our collaborators at the University of California Davis for genetic testing.

The current key was constructed primarily from data collected on museum specimens and specimens collected during disease outbreaks. Consequently, the authors of the key did not have access to data on the coloration of bills, legs, and feet when developing the key. We collected a

series of photos from each bird captured that included key feather tracts, as well as the bill, legs and feet. The tests would help determine the accuracy of the hybrid key when applied in the field and provide additional data that could be used to refine the hybrid key.

4. Conduct field surveys at Hanalei and Hulē‘ia to survey for extent and seasonality of intra- island movements.

Basic methods for this objective are outlined under objective 1. Additionally, we advertised the fact that birds were banded to birding and conservation groups on the islands hoping to increase awareness and possibly effort dedicated to looking for banded Koloa.

5. Develop operational survey to annually detect and remove hybrids from Hanalei and Hulē‘ia NWRs; explore application of these methods to James Campbell NWR on O‘ahu.

This objective is really an outcome that emerges from the data gathered in objectives 2 and 3 and our ability to safely capture large numbers of Koloa. We provide thoughts on this objective in the discussion.

RESULTS

1. Investigate the value of using banded birds and a mark-resight analysis approach to estimate the population size of Koloa using Hanalei and Hulē‘ia NWRs.

Ability to capture and mark birds.-- We trapped Koloa for 312 trap days during three distinct periods (1 Dec 2010 – 24 August 2011, 13 May – 10 June 2012, and 21 November – 14 December 2012). A trap-day was defined as one trap being open for one day. Thus, four trap days could be one trap open for four days or four traps open for one day each. We captured 835 unique Koloa (Table 1). Most birds were captured at Hanalei NWR (97.0%) where the majority of trapping activities occurred (96.5% of trap days). Capture rates were relatively constant among trapping periods. Males (78.4%) were captured in much greater proportion than females (21.6%). We caught 97 individuals of non-target species including 68 (70.1%) Hawaiian

gallinules, 12 (12.4%) Hawaiian coots, 9 (9.3%) Hawaiian geese, 5 (5.2%) greater scaup (Aythya marila), 2 (2.1%) black-crowned night heron (Nycticorax nycticorax), and 1 (1.0%) Hawaiian black-necked stilt.

Future trapping requires that we keep the risk of trapping related injury and mortality low. During trapping, we recorded four trap-related Koloa injuries and two non-target (Hawaiian gallinule) mortalities. The injury rate for Koloa was 0.23%, and the overall injury/mortality rate for all birds captured was 0.32%. In addition, two Koloa eggs were laid in crates while birds waited processing. Trapping ceased following each trap-related injury and mortality so we could submit a report and consult with USFWS permit officials. In addition, we inspected and re- evaluated trap design and protocols and made modifications to reduce the risk of injury/mortality to both target and non-target birds. Trap design modifications included:

- Routinely checked funnel welded wire edges and ensured all edges/points were filed down smooth. - Routinely checked and cable tied gaps between fish mesh and welded wire, as well as interior corners of fish mesh. - Routinely replaced broken or weathered cable ties. - Routinely checked that all cable ties faced outward. - Attached bird netting to upper corner of funnel and padded corners. - Cable tied bird net to top edge of welded wire, and routinely checked and closed gaps. - Locked door flush with interior edges to minimize gaps along interior door edges. - Filled gaps between trap and wetland bottom daily and ensured trap is buried 2 in. - To minimize possible injuries and to exclude Hawaiian goose from entering traps, we attached plastic hardware net to the upper funnel entrance. - Created small islands in traps using mud or logs to provide a place for birds to sit out of water.

Trapping protocol modifications included: - If a trap consistently catches a non-target species it was moved to a new site. - Only trained biologists will approach and remove ducks from traps.

- Only 2-3 biologists approached trap, working quietly and efficiently and avoiding sudden movements (Biologist 1: carried duck crate, closed funnel, placed ducks in crate; Biologist 2: opened door and entered trap as quickly and carefully as possible, removed ducks). - The crew was kept small (3 was optimal). Other crew remained at vehicle. - To reduce the length of time a bird might be in a trap, three trap checks were made daily: the first trap check beginning no more than 1 hour after sunrise, the second no more than 2 hours before sunset, and a final trap check by spotting scope shortly before sunset. Eventually, we modified the protocol so that active traps were checked only during early morning (within one hour of sunrise) when birds were removed from the trap and processed. The trap was then closed (inactive) until just prior to sunset when we baited and reopened the trap

Table 1. A summary of the number of Koloa captured during trapping efforts at Hanalei and Hulē‘ia National Wildlife Refuges from December 2010 to December 2012. New Koloa captures

Koloa Number of Catch per Trapping season Male Female Total recaptures trap days trap daya Dec 2010 – Aug 2011 381 113 494 430 215 2.3 May - June 2012 117 24 141 176 29 4.9 Nov - Dec 2012 157 43 200 312 68 2.9 Total 655 180 835 918 312 2.7 a Calculated as (Total new captures)/(Number of trap days).

Band retention and wear.-- We recorded 918 recaptures during bait trapping. We detected no instance of band loss and no incidence of excessive band wear for either the metal USGS bands or the color bands. A small percentage of bands became discolored by the accumulation of an oxidized compound from the soil; however, the alpha numeric code was still readable.

Ability to locate and read bands in the field or recapture banded birds in traps.-- We conducted two periods of band resighting on the refuge, one in 2011 and another in 2013. Our persistent banding increased the average number of banded birds detected on surveys from 6.2 in 2011 to 34.5 in 2013 (Table 2). However, for most birds it was not possible to determine banded status during surveys (Table 2). Of those birds that were determined to be banded, we were able to read the alpha numeric code for just over 39% of bands in 2011 and 52% in 2013.

Table 2. Summary of band resightings during periodic surveys on Hanalei National Wildlife Refuge, Kaua‘i. Each bird observed was recorded as banded, unbanded, or unknown (its band status could not be determined). Status of Bird Percent # Bands Survey Date Banded Unbanded Unknown Bandeda Read % Readb 02/01/11 8 111 132 6.7 1 12.5 03/04/11 6 144 236 4.0 5 83.3 03/23/11 7 110 216 6.0 4 57.1 04/14/11 7 86 219 7.5 3 42.9 05/20/11 12 63 364 16.0 1 8.3 06/12/11 7 61 271 10.3 3 42.9 06/23/11 3 33 183 8.3 2 66.7 07/06/11 1 15 115 6.3 0 0.0 08/04/11 5 44 297 10.2 2 40.0 2011 Mean (SE) 6.2(1.0) 74.1(14.0) 226(26) 8.4(1.2) 2.3(0.5) 39.3(9.3) 01/31/13 38 98 310 27.9 18 47.4 02/27/13 42 112 349 27.3 16 38.1 04/05/13 25 72 336 25.8 18 72.0 05/05/13 33 89 448 27.0 17 51.5 2013 Mean (SE) 34.5(3.7) 92.8(8.4) 361(30) 27.0(0.5) 17.3(0.5) 52.2(10.2) a Calculated as the percent known to be banded divided by the number of birds known to be unbanded. b Calculated as the number of bands read divided by the number of birds known to be banded.

2. Determine the prevalence of Mallard/Koloa hybrids on Hanalei and Hulē‘ia NWRs.

Between December 2010 and December 2012, we applied the hybrid key to all captured birds (n = 844); nine birds (1.1%) were classified as hybrids and euthanized. No Mallards were captured at either refuge. One Mallard or Mallard-like hybrid was observed, but not captured, at Hanalei NWR on 3/22/11 and 3/30/11. All birds classified as hybrids were male. Four hybrids were identified under Tier 1 of the key (plumage), three under Tier 2 (culmen width ≥ 20.3 mm), and two under both Tiers. One bird that was banded on 1/22/11 as a juvenile with a culmen width of 20.0 mm was recaptured on 3/6/11 with a culmen width of 20.4 mm and subsequently keyed out as a hybrid under Tier 2 criteria (culmen width ≥ 20.3 mm). The nine euthanized hybrids were frozen and stored at Hanalei NWR.

Tissue samples for all euthanized birds were collected and submitted to collaborators at the University of California at Davis and Wright State University for genetic testing to determine the accuracy of the morphological key. Eight of the nine birds were assigned greater than 95% probability Koloa, one bird was <95% Koloa (i.e., scored as “hybrid” based on established criteria) were determined to be Koloa based on genetic results.

3. Collect data to help refine the Mallard/Koloa hybrid key to improve implementation of hybrid removal activities at both Kaua‘i refuges and James Campbell NWR on O‘ahu.

Morphometrics and plumage characteristics were recorded and plumage photographs were taken for all birds captured. In 2010-11 feather samples collected from all birds (n = 495) and blood samples were collected from 305 birds. All collected tissues were shipped to collaborators at the University of California at Davis for genetic analysis. Results of a population genetics study that included a sample of our captured birds from Kauaʻi has since been completed (Wells et al. 2016).

4. Conduct field surveys to characterize the extent and seasonality of intra-island movements.

We conducted seven monthly surveys around Kaua‘i (excluding Hanalei NWR) from February to August 2011. We recorded 481 Koloa during the surveys. Leg status (banded vs not banded) was determined for 183 (38.0%) of birds, and only 6 (3.4%) of those Koloa were banded. We did not detect a bird banded at Hanalei NWR during any of our island-wide surveys except in the lower Hanalei Valley and Lumahai estuary, and we did not resight a bird banded at Hule‘ia NWR outside of that refuge. Since August of 2011 we have detected one bird banded at Hanalei near Hule‘ia and 1-3 birds banded at Hule‘ia at Hanalei.

5. Develop operational survey to annually detect and remove hybrids from Hanalei and Hulē‘ia NWRs; explore application of these methods to James Campbell NWR on O‘ahu.

See discussion.

DISCUSSION

1. Investigate the value of using banded birds and a mark-resight analysis approach to estimate the population size of Koloa using Hanalei and Hulē‘ia NWRs.

We have convincingly demonstrated that a large number of Koloa can be safely captured using a traditional trapping technique developed for waterfowl, the baited swim-in trap. We captured birds throughout the year with consistently high capture, and recapture rates. We did not place traps in taro as that interfered with farming operations; trapping only in managed wetlands is only representative of the population of birds using Hanalei NWR if bird use of the refuge is not habitat specific. In support of that assumption, birds marked with radio transmitters extensively used both managed wetlands and taro (Malachowski unpubl. data). More generally, banded birds and birds affixed with radio transmitters were sighted around Kauaʻi suggesting that trapping at Hanalei might allow inference to the entire island.

Colored alpha numeric bands were functional as auxiliary markers on Koloa. Recaptures of banded birds over a period of several years found no evidence for band loss, either metal or alpha numeric. Some bands were stained extensively, but the codes were still readable when the band was visible through a scope. However, considerable survey effort indicated that reading the alpha numeric color bands was difficult in the field. Our success at reading bands was largely due to the skill and patience of one observer (C. Malachowski) and his investment of considerable time into reading bands. Our approach to resighting will not work for the surveyor who needs to efficiently cover many sites in a single day or for refuge biologists who do not have the time available for band searching with such intensity.

Our inability to efficiently read bands in the field indicates that mark-recapture is the preferable approach for using banded birds to estimate population size. As the name indicates, mark- recapture requires recapturing a sample of marked birds at some future time. Either approach requires a consistent capture and banding effort to maintain a marked sample of birds in the population. The distinctions between the two approaches are the types of marks applied to birds, and depending on the modeling approach being used, the frequency of trapping. Using mark- recapture, birds only need be marked with a metal USGS band, which lowers the costs of marking a population of birds. Assuming a closed population, an instantaneous estimate of population size can be generated from a single trapping action each year that would last 7-10 days. Using this approach, we could assume a closed population and use a simple Lincoln- Peterson estimator to estimate the population size of Koloa using Hanalei NWR at that moment in time.

Alternately, if the capturing protocol included both a primary trapping period (e.g., yearly intervals) and multiple secondary trapping periods (e.g., three trapping periods per primary trapping occasion), population size could be estimated using a “robust design” approach (Kendall and Nichols 1995, Kendall et al. 1995, 1997). The benefits of the robust design are several. First, the longer time between primary capture periods allows for movement of birds onto and off of the refuge and considers temporary immigration, it also allows for several estimates of abundance over the primary trapping interval. Given birds are moving around on Kauaʻi (Malachowski unpubl. data) and moving between Niʻihau and Kauaʻi, trapping at Hanalei

NWR using a robust design would provide one approach for estimating the population size of Koloa on Kauaʻi. Second, a robust design also allows for estimation of survival.

Finally, a monitoring program that incorporates a mark-recapture banding protocol could contribute to decision making regarding Koloa listing decisions. Downlisting and delisting criteria for Koloa include a goal based on population size and trend. For example, the downlisting criteria are “The statewide Koloa population shows a stable or increasing trend at a number greater than 2,000 birds for at least 5 consecutive years.” (USFWS 2011). However, there is currently no estimate of Koloa population size for anywhere in Hawaiʻi. The accuracy of survey results for the statewide waterbird survey is unknown for Koloa, but that count is an index, not an estimate of population size, and no effort has been made to understand the possible biases associated with using that survey to drive recovery decisions. Estimates of population size on O‘ahu and Kaua‘i were thought to be further complicated by the prevalence of Koloa- Mallard hybrids. While that problem remains on Oʻahu, genetic data indicate that the population of Koloa is, currently, largely free of hybrids (Wells et al. 2016); therefore, conditions exist to implement a mark recapture monitoring protocol at Hanalei NWR if managers desire a true estimate of population size.

2. Determine the prevalence of Mallard/Koloa hybrids on Hanalei and Hulē‘ia NWRs.

We detected few birds that keyed as hybrids during trapping. This suggests the percent of hybrids on Kaua‘i is lower than previously believed or that the key we used had low power to detect hybrids. Subsequent genetic testing of a sample of birds captured revealed no significant evidence for hybrid presence at Hanalei (Wells et al. 2016), supporting our results based on plumage and morphometrics. Previous estimates for the prevalence of hybrids on the island have been based on observations of birds with mallard plumage traits seen during periodic surveys. Our work on plumages and molts in conjunction with collaborators at the University of California Davis has revealed significantly more variation in male plumage than previously described for Koloa. This variation is known to occur in birds that are genetically confirmed to be Koloa (Wells et al. 2016); thus, any estimate of hybrid prevalence based solely on visual sightings will likely be biased high for males.

3. Collect data to help refine the Mallard/Koloa hybrid key to improve implementation of hybrid removal activities at both Kaua‘i refuges and James Campbell NWR on O‘ahu.

The large number of Koloa captured in this study allowed us to document a more complete array of male plumage variation than was previously described (Engilis pers. comm.). Additionally, photos for each bird documented considerable variation in bill, leg and foot color, features that were not available for consideration while developing the initial hybrid key, which was built on data collected from museum specimens. During 2012, we met with A. Engilis (co-author of hybrid key) to review some of our photo data and discuss ways such information might be incorporated into future versions of the key.

In addition, these data will help validate and improve the existing morphological key used for identifying hybrids in the field (e.g., Wells et al. 2016), as well as contribute to our understanding of the range of age-specific plumage variation in non-hybridized Koloa.

The results of genetic testing determined the accuracy of using the hybrid key in the field and provide us with an unbiased estimate of the percent of Koloa using Hanalei that are hybrids. Data collected during this project revealed considerable variability in plumage patterns not seen in the population of birds used to develop the initial hybrid key. That information, combined with genetic testing, may identify additional plumage traits useful for identifying Mallard/Koloa hybrids. Further, recapturing birds in 2012 that had gone through one year of molts will greatly increase our understanding of variation in feather color and patterning that should be useful for improving the morphology key.

Continued use of the existing morphometric key would result in false positives for 1% of birds that score as “hybrids”. Given examples of variation in culmen measurements among measurers and among attempts for a single measurer on a single bird, a more risk-averse method of using the morphometric key would involve building in measurement error to the decision making threshold criteria values. For example, if the measurement error for culmen was ± 0.3 mm, and a hybrid is defined as any bird with a culmen wider than 20.3 mm, the hybrid identifying threshold

could be increased to 20.6 mm. Importantly, the ability of the existing morphometric key to identify hybrids is unknown, and remains to be tested, as it has not been applied to a population of known hybrids. The new genetic results from islands other than Kaua‘i provide a sample of birds to conduct that test. We recommend that such a comparison be made as a means of completing validation of the existing morphometric data.

4. Conduct field surveys at Hanalei and Hulē‘ia to survey for extent and seasonality of intra- island movements.

The ongoing bi-annual, statewide and monthly Refuge waterbird surveys will provide additional opportunities to resight banded birds around Kaua’i. However, since banding began in 2010, no Kaua‘i-banded birds have been resighted on Oʻahu. However, Kauaʻi banded birds have been resighted during refuge surveys and state surveys (in and around HANWR).Our work at Hanalei indicates that observing bands requires careful observation at fairly close range, which may not be possible at many sites or during surveys when a large number of sites must be covered in a day. Subsequent work on birds affixed with VHF transmitters (Malachowski unpubl. data) indicates that birds do move back and forth among locations like Hanalei, Hulēʻia, and the Mana Plain. Given the difficulty of observing banded birds, some form of telemetry (e.g., VHF, GPS) is the best tool for answering question about movements of birds within and among islands.

Extensive survey activities on Oʻahu during November 2013 and June–July 2015 failed to locate any marked birds from Kauaʻi. Despite banding more than 1,000 Koloa on Kaua‘i since 2010, surveys on O‘ahu have failed to find any banded birds from Kaua‘i. Additionally, no birds fitted with VHF transmitters or PTTs moved from Kaua‘i to O‘ahu despite the fact that movements between Ni‘ihau and Kaua‘i were relatively common for some birds (Dugger and Malachowski 2013). Together, these lines of evidence suggest movements by birds from Kaua‘i to O‘ahu are not common; prevailing wind patterns do run contrary to this direction of movement. However, our understanding about the frequency of movements between O‘ahu and Kaua‘i is incomplete because birds on O‘ahu have not been banded or transmittered, so we lack data on the probability of movement from O‘ahu to Kaua‘i. We think additional efforts to estimate that probability are desirable, but genetic analyses completed to date lead us to predict that such movements are rare.

5. Develop operational survey to annually detect and remove hybrids from Hanalei and Hulē‘ia NWRs; explore application of these methods to James Campbell NWR on O‘ahu.

Both genetic and morphological data support the conclusion that Koloa-Mallard hybrids are rare on Kaua‘i; this is good news for species recovery. We recommend periodic, continued efforts to monitor for hybrids at Hanalei NWR and, given the difficulty of identifying hybrids visually using binoculars or spotting scope, we recommend this be accomplished by periodically capturing a sample of birds whose plumage characteristics and measurements can be compared against a revised or validated morphological key. Our work has convincingly demonstrated that baited swim in-traps are effective at safely capturing large numbers of Koloa.

The number of trapping periods per year should be determined by the prevalence of hybrids; given the currently low hybrid prevalence, one trapping period every 1-3 years is likely sufficient. If the objective was to also have a population of marked birds for generating estimates of population size and survival, then once a year is the minimum sampling effort required. When using multiple trapping periods they should be evenly distributed throughout the year to maximize the possibility of capturing birds that may move among sites within a year.

The basic “unit” of activity is a trapping session, defined as the total number of days involved with the setup, operation of, and take down of traps. We recommend a trapping session be defined as three –four weeks that includes two – three weeks of active trapping. If the sole intent of a trapping period is to capture, identify, and remove hybrids then the minimum activities would involve taking several measurements for each captured bird and banding all birds before release. Banding birds that are released would make it easy to identify and release recaptures whose status has already been determined. At Hanalei, one, two-person team can operate 2-3 active traps/day, but having more traps (4-5) available is the most efficient way of maintaining an efficient and constant capture effort because it allows for the preparation of new trap sites while captures are occurring at active traps, and provides greater spatial distribution/coverage on the refuge and likely increases recapture rates (thereby increasing precision of parameter estimates).

One individual should be dedicated to trapping full time during the trapping period while the second individual is needed only for the morning trap checks (2-3 hours). The number of trap days per trap period will depend in part on the population size of Koloa at a site and the size of the site. At Hanalei we found trapping for 2-3 weeks was sufficient to cover most of the trappable areas on the refuge.

In June-July 2015, we conducted a trapping session at James Campbell NWR to assess the utility of bait traps elsewhere on the Hawaiian Islands and collect preliminary data on the prevalence of hybrids on O‘ahu. We carried out a two week trapping session on O‘ahu from 20 June – 5 July 2015. Activities associated with preparation to trap meant that traps were set for the last 4 days of that period at James Campbell.

During 2-5 July we captured 18 individual Koloa or Mallard-Koloa hybrids at James Campbell NWR. The population of Koloa at James Campbell is much lower than Hanalei, thus, fewer birds were available to be captured. Of those birds, 7 were female and 11 were male. Removing ducklings and juvenile birds from the data set, we captured 16 birds, 6 females and 10 males, for a sex ratio of 1.7:1. Application of the morphometrics key to the 10 males indicated that 2 males scored as a hybrid based on culmen width. No males would have scored as hybrids based on plumage characteristics. Mean body mass was 707 g for females and 771 g for males.

Prior to opening traps at James Campbell, we assisted State Division of Forestry and Wildlife (DOFAW) employee Stephen Turnbull to capture birds at sites around O‘ahu outside of James Campbell NWR. Between 20 June and 2 July, we captured 17 birds. We collected data on an additional 8 birds that were recovered dead, likely of botulism. Of the 17 birds we captured, 7 were female and 10 were male. All 10 males captured scored as hybrids based on culmen width and 1 scored as hybrid based on both culmen width and plumage characteristics.

Genetic analyses of these birds plus additional birds from Kauaʻi were completed in December 2016 (Wells et al. 2016). While we refer readers to that report for a complete description of results, those results indicate the Kaua‘i population of Koloa is genetically distinct, while

populations on all other islands show varying levels of hybridization. There was considerable variation in the level of hybridization among subpopulations of birds on O‘ahu, more than we would have predicted based on movement data from Kaua‘i. Our results from banding and movement studies (Malachowski et al. unpubl. data) and those in Wells et al. (2016) provide the foundation for a new conversation to address hybridization concerns for Koloa on Hawaiʻi.

ACKNOWLEDGMENTS

Funding and support for this work has been provided by U.S. Fish and Wildlife Service (Region 1 Inventory and Monitoring Program, PIFWO, Region 1 Migratory Birds, and Invasive Species Program), Pacific Coast Joint Venture, Oregon State University, North Star Science and Technology and American Bird Conservancy, Kīlauea Point Natural History Association, Ducks Unlimited Canada, California Waterfowl Association, Hawai‘i DLNR, KUPU/Americorps. Special thanks to Shannon Smith, Mike Mitchell, and Chadd Smith of Kaua‘i NWRC. There have been a great number of people that have helped us either in the logistics setting up the project or time in the field collecting data on the project; we greatly appreciate their contributions, in alphabetical order this includes: Mike Adams, Tracy Anderson, Linda Belluomini, Chadd Billat, Brad Bortner, Mike Couffer, Megan Dalton, Kristina Dickson, Alex Diego, Sharon Donnelly, John Eadie, Andy Engilis, Mike Fitzgerald Sr., Holly Freifeld, Adam Gaylord, Holly Freifeld, John Gorey, Cody Hagen, Jennifer Hahn, Rodney and Karol Haraguchi, Jamie Harris, Darryl Heard, Adonia Henry, Thomas Ka‘iakapu, Elizabeth Kain, Kevin Kilbride, Marilou Knight, Gary Koga, Mark Koga, Gregory Koob, David Kostersky, Kara Lee, Miguel and Jean Legaspi, Jimmy Macaulay, Lindsay Malachowski, Annie Marshall, Marie McKenzie, Angie Merritt, Daniel Mulcahy, Jay Nelson, Brody Nordland, Emily Pollam, Flora Quick,

Michelle Reynolds, Emily Rider, Lex Riggle, Nanette Seto, Scott Sims, Charlie and Diana Spencer, Bethany Spiegel, Whitney Taylor, Caroline Tucker, Jared Underwood, Julie Unfried, Kathy Valier, Bobby Watari, Caitlin Wells, Keith Wong, and others who volunteered time and expertise.

LITERATURE CITED:

Banko, W. E. 1987. Historical synthesis of recent endemic Hawaiian birds. Koloa-Maoli. University of Hawaii, Manoa, HI, USA.

Bellrose, F. C. 1976. Ducks, geese, and swans of North America. 2nd ed. Stackpole Books, Harrisburg, PA. 544pp.

Browne, R. A., C. R. Griffin, P. R. Chang, M. Hubley, and A. E. Martin. 1993. Genetic- divergence among populations of the Koloa, Laysan duck, and Mallard. Auk 110:49-56.

Devries, J.H., R.W. Brook, D.W. Howerter, and M.G. Anderson. 2008. Effects of Spring Body Condition and Age on Reproduction in Mallards (Anas platyrhynchos). The Auk 125(3):618–628.

Dubois, D. 1975. An evaluation of traps and baits for capturing waterfowl in coastal Louisiana. Thesis, Louisiana State University, Baton Rouge, USA

Ducks Unlimited. 2003. Great Lakes Mallard Study Procedures Manual, 3rd edition. Great Lakes/Atlantic Regional Office, Ann Arbor, MI, USA. 38pp.

Engilis, A., Jr., K. J. Uyehara, and J. G. Giffin. 2002. Koloa (Anas wyvilliana), The Birds of North America Online (A. Poole, Ed.). Ithaca: Cornell Lab of Ornithology; Retrieved from the Birds of North America Online: http://bna.birds.cornell.edu/bna/species/694. Retrieved 5 October 2009.

Engilis, A., Jr., and J. Mc. Eadie. Key to Identify Koloa and Mallard/Koloa Hybrids in the Hawaiian Islands. Oral presentation to U.S. Fish and Wildlife Service and Hawaii Department of Land and Natural Resources Officials. May 20, 2009. Honolulu, Hawaii.

Fair, J., E. Paul, and J. Jones, Eds. 2010. Guidelines to the Use of Wild Birds in Research. Washington, D.C.: Ornithological Council.

Fowler, A. C., J. M. Eadie, and A. Engilis. 2009. Identification of endangered Koloa (Anas wyvilliana), introduced North American Mallards (A. platrhynchos) and their hybrids using multilocus genotypes. Conservation Genetics. 10:1747-1758.

Giffin, J. G. 1983. Abundance and distribution of Koloa on the Island of Hawaii: movements, survival, reproductive success and habitat of Koloa on the Island of Hawaii. Pittman- Robertson report (W-18-R-7, W-19-R-8, R-III-H). Hawaii Div. For. Wildl., Honolulu, HI, USA

Gustafson, M. E., J. Hildenbrand and L. Metras. 1997. The North American Bird Banding Manual (Electronic Version). Version 1.0

Harder, J. D. and R. L. Kirkpatrick. 1996. Physiological methods in wildlife research. Pages 275- 306 in T. A. Bookhout, editor. Research and management techniques for wildlife and habitats. The Wildlife Society, Bethesda, MD, USA.

HCA [Hawai‘i Conservation Alliance]. 2008. Position Paper: Feral Mallards in Hawaii. Prepared by Uyehara, K. J., Engilis, A. E., Marshall, A., Koob, G., Henry, A., and B. Zaun.

Heard, D. J., D. M. Mulcahy, S. A. Iverson, D. J. Rizzolo, E. C. Greiner, J. Hall, H. Ip, and D. Esler. 2008. A blood survey of elements, viral antibodies & hemoparasites in wintering harlequin ducks (Histrionicus histrionicus) and Barrow’s goldeneyes (Bucephala islandica). Journal of Wildlife Diseases 44:486-493.

Henshaw, H. W. 1902. Birds of the Hawaiian Islands. Thos. G. Thrum, Honolulu, HI, USA.

Hoekman, S.T, T.S. Gavor, R. Maher, H.R. Murkin, M.S. Lindberg. 2006. Demographics of Breeding Female Mallards in Southern Ontario, Canada. Journal of Wildlife Management, 70.

Hunt, G. S., and K. J. Dahlka. 1953. Live trapping of diving ducks. Journal of Wildlife Management 17:92-97.

Kendall, W. L., and J. D. Nichols. 1995. On the use of secondary capture-recapture samples to estimate temporary emigration and breeding proportions. Journal of Applied Statistics, 22:751-762.

Kendall, W. L., K. H. Pollock, and C. Brownie. 1995. A likelihood-based approach to capture- recapture estimation of demographic parameters under the robust design. Biometrics 51: 293-308.

Kendall, W. L., J. D. Nichols, and J. E. Hines. 1997. Estimating temporary emigration using capture-recapture data with Pollock’s robust design. Ecology 78:563-578.

Korschgen, C. E., K. P. Kenow, A. Gendron-Fitzpatrick, W. L. Green, and F. J. Dein. 1996. Implanting intra-abdominal radiotransmitters with external whip antennas in ducks. Journal of Wildlife Management 60:132-137.

Moulton, D. W. and M. W. Weller. 1984. Biology and conservation of the Laysan Duck (Anas laysanensis). Condor 86:105-117.

Paulus, S. L. 1984. Behavioral ecology of mottled ducks in Louisiana. Ph.D. dissertation, Auburn University, Auburn, Alabama, USA. 152pp.

Olson, S. L., and H. F. James. 1982. Prodromus of the fossil avifauna of the Hawaiian Islands. Smithsonian Contributions to Zoology 365:1-59.

Perkins, R. C. L. 1903. Fauna hawaiiensis: Vertebrata (Aves). Columbia University Press, New York, NY, USA.

Ramsey, F.L., and D.W. Schafer. 2002. The Statistical Sleuth, A Course in Methods of Data Analysis, 2nd edition. Duxbury Press, Pacific Grove, California, USA.

Reynolds, M. H. 2002. The foraging ecology, habitat use, and population dynamics of the Laysan teal (Anas laysanensis). Ph.D. dissertation. Virginia Polytechnic Institute and State University, Blacksburg, Virginia, USA. 187pp.

Rhymer, J. M. 2001. Evolutionary relationships and conservation of the Hawaiian anatids. Studies in Avian Biology 22:61-67.

Schwartz, C. W., and E. R. Schwartz. March 1953. Notes on the Koloa. The Wilson Bulletin 65:8.

Swedberg, G. E. 1967. The Koloa: a preliminary report on the life history and status of the Koloa (Anas wyvilliana). Dep. Land Nat. Resour., Honolulu, HI, USA.

U.S. Fish and Wildlife Service. 2011. Recovery Plan for Hawaiian Waterbirds, Second Revision. U.S. Fish and Wildlife Service, Portland, Oregon. xx + 233 pp.

Uyehara, K. 2007. Site Reconnaissance for Koloa (Koloa) Study, May 2006: U.S. Geological Survey Open File Report 2007-1351, Reston, Virginia.

Uyehara, K., Engilis, A, Marshall, A., Koob, G., Henry, A., and B. Zaun. 2008. The Feral Mallard Threat to Hawaii’s Native Duck: Recommendations for Preventing the Extinction of the Endangered Koloa Maoli. Hawaiʻi Conservation Alliance.

Weller, M. W. 1956. A simple field candler for waterfowl eggs. Journal of Wildlife Management 20:111-113.

Wells, C.P., S. Turnbull, P. Lavretsky, K. Uyehara, C. Malachowski, B. D. Dugger, J. M. Eadie, and A. Engilis, Jr. 2016. Determining level of hybridization between Koloas (Anas wyviliiana) and Mallards (A. platyrhynchos). Project Report. Hawaii Division of Forestry and Wildlife.

Wobeser GA. 1997. Diseases of Wild Waterfowl. 2nd. Ed. Plenum Press, New York.