Mammal-exclusion fencing improves the nesting success of an endangered native Hawaiian waterbird

Introduction

Endangered species often face multiple threats, leaving decision-makers with the task of discerning which set of actions is most likely to achieve recovery, given budget constraints (Beyer et al., 2018; Carwardine et al., 2019). Further, potential actions to abate a given threat often vary in effectiveness and cost resulting in further decision complexity (Tulloch et al , 2015). Managers may improve outcomes for future decisions by monitoring the response of protected species following conservation interventions (Runge, 2011). However, few studies utilize monitoring data to implement adaptive management in a structured manner (Williams, Balmford & Wilcove, 2020). A critical evaluation of predictions made prior to taking action, in light of actual outcomes, can improve our understanding regarding how a system works (Moon et al., 2019). This process is particularly important when there are multiple options available to abate a given threat, each with varying costs and impacts.

Invasive predators, particularly mammals, contribute to native bird declines and species extinctions worldwide (Doherty et al., 2016). Ground-nesting birds, in particular, tend to benefit the most from control of these predators (Holt et al., 2008; Lavers, Wilcox & Donlan, 2010). The task of managing invasive mammalian predators is currently approached by poison-baiting, trapping and removal, and/or exclusion fencing (O’Donnell, Clapperton & Monks, 2014). Poison-baiting is the most cost-effective method of predator removal over a large area, but this option may impact non-target species and often carries a high social cost (Howald et al., 2005; Duron, Shiels & Vidal, 2017). Trapping and removal, as well as exclusion fencing, minimize secondary impacts to non-target species. However, trapping and removal efforts may be ineffective if there is increased predator fecundity or survival, rapid immigration from adjacent sites, or if too few predators are removed to have a detectable benefit (Côté & Sutherland, 1997; Meckstroth & Miles, 2005). Furthermore, removal of a single predator type, such as feral cats (Felis catus), may lead to the release of meso-predators, such as rats (Rattus spp.), thereby increasing predation rates (Rayner, Hauber & Clout , 2007). Thus, exclusion of a diversity of predator types by exclusion fencing may be more effective in protecting vulnerable species (Smith et al., 2010).

Exclusion fences serve an important role in supporting the recovery of some species (Smith et al. , 2011; Innes et al., 2012) including passerines, seabirds, and shorebirds (Neuman et al., 2004; Sanders, Brown & Keedwell, 2007; Rickenbach et al., 2011; Malpas et al. , 2013; Young et al., 2013). Successful implementation is heavily influenced by design and placement. Managers may strategically use natural barriers such as cliffs, peninsulas, and bodies of water to reduce costs and improve results. However, these designs may be more susceptible to mammalian incursions as these designs are not fully encircled (Young et al. , 2012). Additionally, enclosures that use bodies of water as a natural barrier may still be accessible to amphibious and aquatic predators, and most enclosure designs are vulnerable to avian predators. While many studies have evaluated different predator control methods, no study to date has evaluated the effectiveness of exclusion fencing in protecting endangered waterbirds nesting in coastal environments.

The Hawaiian Stilt (Ae‘o, Himantopus mexicanus knudseni), a sub-species of the Black-necked Stilt (Himantopus mexicanus), declined in the 19th century due to habitat loss and degradation, over-harvesting, and the introduction of invasive predators (USFWS, 2011). Following listing as endangered by the U.S. Fish and Wildlife Service (USFWS) in 1970, state and federal efforts have established over 6,000 hectares of designated habitat (Reed et al., 2011; USFWS, 2011). Predator removal efforts vary by site and re-incursions by feral cats, rats, and mongoose (Herpestes javanicus) from nearby areas are common. Indeed, introduced mammalian predators remain a major threat to Hawaiian Stilt nesting success, even within state and federally managed wetlands (Harmon Wehr & Price, 2020). In 2011 the population was estimated at ∼1,500 individuals, with a slowly increasing trend (Reed et al., 2011; Reed et al, 2015). However, habitat availability may be a limiting factor in achieving recovery (Reed, Elphick & Oring , 1998) and is likely to decrease with sea level rise (Harmon et al. 2020b). Thus, it is particularly important that reproductive success in available habitat is maximized.

Prior to the introduction of invasive predators, native predators of the Hawaiian Stilt were limited to avian species such as the Hawaiian Short-eared Owl (Pueo; Asio flammeus sandwichensis), the Hawaiian Hawk (I‘o; Buteo solitarius), the now extinct Stilt Owl (Grallistrix spp.), as well as other native waterbirds such as the Black-crowned Night Heron (Auku‘u; Nycticorax nycticorax). Today, a plethora of non-native predators from a variety of taxonomic groups prey on adult stilts, chicks, and eggs (Underwood, 2013; Harmon Wehr & Price, 2020). Avian predators predominantly use visual cues to locate potential prey, whereas nocturnal predators, such as mammals, may rely on olfaction (Conover, 2007). Hawaiian Stilts have developed a few anti-predator behaviors such as broken wing displays and aggressive “dive-bombing.” Although these behaviors are largely efficient in deterring avian predators (Sordahl, 2004), Hawaiian Stilts also remove eggshells from the nest after chicks have hatched to reduce scent, an antipredator behavior typical of North American Recurvirostrids (Sordahl, 1994). Soon after the brood hatches, adults lead their young away from the nesting area (Colemann, 1981), possibly in search of better foraging or to evade a lingering scent that may attract potential predators, or both. Further, Hawaiian Stilts have the capacity to re-nest, combine egg clutches, and aggressively respond to potential predators, suggesting they may be resilient against a variety of predators, similar to Black-necked Stilts (Colemann, 1981; Sordahl, 2004).

In this study we compared the nesting success of Hawaiian Stilts in a wetland protected by a mammal-exclusion fence and a regular trapping regime to a nearby wetland without an exclusion fence, where mammals were removed solely via trapping. We predicted nesting success would be greater for breeding pairs inside the exclusion fence and the hatchlings would spend more time in the nesting area compared to the site where mammals were controlled via trapping and removal only.

Methods

Study area

This study was conducted at two wetland units, Honouliuli and Waiawa (Fig. 1), both part of the larger Pearl Harbor National Wildlife Refuge (PHNWR) complex. PHNWR was established in 1976 as mitigation for the construction of the Honolulu International Airport Reef Runway (USFWS, 2009). In August 2018, the USFWS completed a 1,006 m mammal-exclusion fence (Fig. 2) around the Honouliuli wetland unit (Fig. 3), in which one side is open to an estuary (Fig. 4) and is based on the Xcluder™ Kiwi model designed for multiple-species exclusion (Day & MacGibbon, 2007). The two sites are in the same region on the island of O‘ahu with similar proximity to urban development and similar rainfall regimes ranging from 555–619 mm annually (Giambelluca, Shuai & Businger, 2014). The Honouliuli unit (hereafter referred to as ‘fenced site’; 14.7 hectares) is typically less saline than the Waiawa unit (hereafter referred to as ‘unfenced site’; 9.9 hectares; Colemann, 1981). Plant community composition differs somewhat between sites with water hyssop (‘Ae‘ae ; Bacopa monnieri), non-native pickleweed (Batis maritima), and cattails (Typha spp.) dominating the fenced site, while the unfenced site is dominated almost exclusively by non-native pickleweed and Guinea grass (Megathyrus maximus) along the perimeter.

Table 1:

Potential predators of the Hawaiian Stilt and the life stage they depredate.

Common Name	Scientific Name	Eggs	Chicks	Reference
Barn Owla	Tydo alba		x	Robinson et al. (1999)
Black-crowned Night Heron	Nycticorax nycticorax	x	x	Colemann (1981)
Bullfroga	Rana catesbeiana		x	USFWS (2009)
Cattle Egreta	Bubulcus ibis	x	x	Colemann (1981)
Common Mynaa	Acridotheres tristis	x	x	Colemann (1981)
Domestic Catab	Felis catus	x	x	Colemann (1981)
Domestic Doga	Canis familiaris	x	x	Colemann (1981)
Hawaiian Short-eared Owl	Asio flammeus sandwichensis	x	x	Colemann (1981)
Indian Mongoosea	Herpestes javanicus	x	x	Colemann (1981)
Large fish	–		x	USFWS (2011)
Laughing Gull	Larus atricilla	x		Colemann (1981)
Piga	Sus scrofa	x		Underwood et al., (2014)
Ratab	Rattus spp.	x		Colemann (1981)
Ruddy Turnstone	Arenaria interpres	x		Colemann (1981)

DOI: 10.7717/peerj.10722/table-1

Notes:

aIndicates a non-native introduced species.

bIndicates potential predators confirmed in this study.

Both sites are managed by the USFWS, with similar predator communities prior to the construction of the fence (Table 1). Trapping by USFWS personnel occurred regularly at both sites between 2018 and 2020. USFWS used Tomahawk style live traps meant to target rats, feral cats, and mongoose as well as DOC 250 traps that primarily target rats and mongoose. The live traps were opened on Mondays, checked on Wednesdays and closed on Fridays. Traps were baited with Koi food, cat food, salmon oil, sardines, Vienna sausage, and/or clam juice and targeted rats, feral cats, mongoose, and mice. The unfenced site had a combination of 16 DOC 250 traps and 13 live traps while the fenced site had nine DOC 250 traps and seven live traps (Figs. S1 & S2).

Results

At the unfenced site, 37 nests were discovered (3.73 nests/ha), compared with 18 total nests discovered inside the mammal-exclusion fence (1.22 nests/ha). Of the discovered nests, 21 of 37 (57%) at the unfenced site, and 9 of 18 (50%) at the fenced site, had useable camera data that spanned egg laying through hatching, and therefore were included as ‘observed nests’ in this study. The remaining nests were either discovered too late in the nesting cycle, or suffered a battery failure/memory malfunction resulting in a lack of data and exclusion from analyses.

Significantly more eggs were laid per nest at the fenced site (4.00 ±  0.00; t₂₀ = 3.16, P = 0.005), compared with the unfenced site (3.67 ±  0.11). At both sites, the number of eggs hatched per nest was significantly less than the number of eggs laid (fenced: t₈ = 2.31, P = 0.050; unfenced: t₂₀ = 7.63, P < 0.001); however, a significantly greater number of eggs hatched per nest at the fenced site (3.33 ±  0.29; t₂₅ = 4.71, P <  0.001; Fig. 5) compared to the unfenced site (1.29 ±  0.33). At the fenced site, none of the observed nests were depredated and all nests had at least one egg hatch (n = 9). Further, five of the nine nests (55.6%) hatched the full clutch. At the unfenced site, 10 of 21 nests (47.6%) had at least one egg hatch and six of 21 (28.6%) were depredated, three by a feral cat, two by rats and the sixth fell prey to an unknown predator. Five nests (23.8%) were abandoned, flooded, or had unknown fates (Fig. 6). Three of 21 (14.3%) nests at the unfenced site hatched the full clutch . At the unfenced site, clutch sizes did not differ between early (3.70 ±  0.15, n = 10) and late nesters (3.6 ±  0.16, n = 10; t ₁₈ = 0.45, P = 0.66).

Seven nests at the fenced site and nine nests at the unfenced site had usable data to test time spent by chicks in the nesting area following hatching. There was no difference between sites in the time spent by the stilt chicks in the nesting area after hatching (t₁₄ = 0.59, P = 0.56; Fig. S5). During the 2019 study period, potential predators detected by our cameras at the fenced site included the Black-crowned Night Heron, Cattle Egrets (Bubulcus ibis), Common Mynas (Acridotheres tristis), and Bullfrogs (Rana catesbeiana). At the unfenced site, in addition to the species observed at the fenced site, potential predators detected included rats, feral cats, and the small Indian mongoose. Defensive behaviors were observed at each site toward bullfrogs, Black-crowned Night Herons, and Common Mynas. Defensive behaviors included barking, mobbing, dive-bombing, and aerial chasing.

Inside the fenced site between 2018 and 2020 only one feral cat was trapped. Meanwhile, 26 feral cats were trapped within the unfenced site during the same time period (Table 2). Both rats and mongoose were detected and/or trapped at both sites from 2018 to 2020 (Figs. S6–S8).

Discussion

As Hawaiian waterbird nesting habitat is likely to decline due to sea level rise (Harmon et al., 2020b), conservation actions maximizing reproductive success in remaining habitat are necessary to achieve recovery. Prior to this study, the effectiveness of mammal-exclusion fencing versus trapping of mammals alone had not been evaluated. Within the first nesting season after construction of a mammal-exclusion fence at a federally protected wetland, we found that nesting success was significantly higher for Hawaiian Stilts nesting inside the fence, compared with those nesting in a nearby wetland where trapping and removal of mammalian predators was the sole means of predator control. Unexpectedly, a significantly higher number of eggs were laid per nest at the fenced site compared with the unfenced site. Together, this resulted in nearly three times the number of eggs hatched per nest inside the mammal-exclusion fence, compared with nests at the site with trapping alone. Thus, mammal-exclusion fencing may play an important role in maximizing reproductive success for Hawaiian waterbirds, as sea level rise decreases available nesting habitat.

Importantly, our results suggest that mammalian predators are the greatest threat to hatching success in Hawaiian Stilts, as the complete exclusion of mammals from the nesting area resulted in substantial gains in the number of eggs laid and chicks hatched per nesting pair. Potential predators such as herons, owls, crabs, turtles, and bullfrogs were not excluded by the fence, and thus were expected to potentially impact nesting success at both sites. Despite detection of these predator types inside the mammal-exclusion fence, they were not observed depredating eggs, potentially due to defensive behaviors such as mobbing, dive-bombing, or others typical of Black-necked Stilts (Sordahl, 2004). Anecdotally, on several occasions, stilt pairs were observed engaging with Black-crowned Night Herons well out of the frame of the camera. Stilts may engage with different predators at different distances (Sordahl, 2004) and any interactions out of the frame of the camera could not be recorded consistently. Further, as detections by motion-activated cameras are imperfect, cameras may have missed predation events or visits to the nests by potential predators.

This study did not track fledging success of the stilts and potential predators may have impacted mortality of chicks at a later stage, hindering recruitment. Further research is needed that evaluates the impact of exclusion fencing on fledging success in Hawaiian Stilts, as well as parental and chick behaviors inside and outside of mammal-exclusion fencing. Differences in water salinity, food availability, and other factors not considered here may also have influenced nesting success. If other factors do not impact fledging success and juvenile survival prior to recruitment, we may expect increased nesting success to result in increased recruitment and larger breeding populations (Smith et al. , 2011).

Our 2019 camera trap data corroborates the USFWS tracking tunnel and trapping data for the same year. Although USFWS staff visit both sites when checking traps, the unfenced site had more traps deployed between 2018 and 2020 resulting in greater trapping effort or more ‘trap nights’ at the unfenced site, potentially explaining why more feral cats and mongoose were trapped at the unfenced site. Regardless, these data do suggest smaller populations of feral cats and mongoose exist at the fenced site. Future studies are needed to compare the cost-effectiveness of different trapping levels with that of exclusion fencing.

Mammal-exclusion fencing may decrease physiological stress in nesting birds induced by the need to remain vigilant for predators, thereby increasing reproductive success. A decreased need for vigilance has been associated with larger clutch sizes and increased fitness in other species (Zanette et al., 2011; Thomson et al., 2010). This is consistent with our observation of larger clutch sizes for breeding pairs nesting inside the mammal-exclusion fence. While smaller clutch sizes can result from parents renesting after a failed attempt earlier in the season, we found no difference in clutch sizes at the unfenced site between early and late nesters, suggesting that the smaller average clutch size at the unfenced site compared to the fenced site is not likely due to renesting individuals. However, because most adult stilts were not banded, we were unable to identify if and when individuals renested following a failed nesting attempt. Furthermore, between the sites, there was no detectable difference in the time spent in the nesting area by stilt chicks following hatching. Adult stilts typically lead their broods away from the nest to nearby foraging areas within several days of hatching (Colemann, 1981), potentially to avoid predators that may have discovered the nest site.

When considering the cost effectiveness of mammal-exclusion fencing, conservation planners may also wish to consider the number of species protected, the number of individuals per species protected, and potential gains in reproductive success across species, as well as the trapping effort required to achieve similar gains in reproductive success. The exclusion fence in this study enclosed a relatively small area and protects endangered, native, and migratory species. With a life expectancy of up to 25 years, and a financial break-even point of 16 years compared to trapping alone, fences such as these can be cost-effective over the life of the fence (Moseby & Read, 2006; Sanders, Brown & Keedwell, 2007; Malpas et al. , 2013; Young et al., 2013). The results of this study, demonstrating the substantial reproductive cost of mammalian predators to waterbird nesting success, even with trapping and removal of invasive mammals at the unfenced site, highlight the importance of comparing the effectiveness of potential management actions alongside costs. Future studies should also evaluate the impact of multiple levels of trapping effort at non-fenced sites, to inform comparisons of trapping effort and exclusion fencing, alongside other potential conservation actions (scofield, Ross & Maggie, 2011).

Conclusions

For conservation-reliant species, particularly those on islands where invasive predator eradication is not currently taking place, conservation fencing may provide a means of increasing nesting success. To our knowledge, our study is the first to evaluate the effectiveness of mammal-exclusion fencing in protecting Hawaiian waterbirds during the breeding season. Our results suggest that mammal-exclusion fencing substantially increases nesting success for at least one endangered waterbird. Further, it suggests that invasive mammals may be the greatest predator-based threat to persistence in this species, as their exclusion from the nesting area resulted in a three-fold increase in the number of chicks hatched per nest.

For other endangered waterbird species that breed year-round inside the enclosure, such as the Hawaiian Coot (‘Alae ke‘oke‘o; Fulica alai) and Hawaiian Gallinule (‘Alae ‘ula; Gallinula galeata sandvicensis), the recruitment potential could also be substantial. Gains from exclusion fencing may be greater still at sites adjacent to highly urbanized habitats with large source populations of invasive mammals. Exclusion fencing projects may also capture the interest and support of the public, especially communities in close proximity to the refuge (Innes et al., 2012; Innes et al., 2019). As complete island-wide eradication of mammalian predators is not feasible at this time, and endangered Hawaiian waterbirds are conservation-reliant due to impacts from invasive mammals and the necessity of habitat management for nesting success (Reed et al., 2012; Underwood et al. 2013), our study suggests that mammal-exclusion fencing is a highly effective option to increase nesting success in remaining habitat.

Supplemental Information