Co-breeding involving herons and a potential egg predator,the Indian House Crow (Corvus splendens), in Peninsular India

Association between species may strengthen the fitness of the species involved It is not rare that avian species associate on the breeding and feeding grounds. However, a species associated with a potential egg predator is less common. In this study, a synchronized breeding of Indian House Crow (Corvus splendens) and breeding Indian Pond Herons (Ardeola grayii) in urban conditions is reported. Both the crow abundance and the crow nest abundance increased with the number of heronry nests on sites. Crows were mostly observed when flying over or when resting nearby, but they also attempted egg predation from heronry nests. Crows also used the heronry sites for collecting nesting resources, such as twigs, scavenging dead chicks and for stealing the food brought to feed the heronry chicks. A dearth of suitable nesting places and provisions in an urban environment may be the reason why these birds share nesting trees. Vigilant breeding crows, despite their ability to depredate heron nests, may be more beneficial to herons as they are known to mob and distract heron predators, but a full cost-benefit analysis needs to be undertaken.     

Blue penguin (Eudyptula minor) nest distribution and breeding success on Otago Peninsula, 1992 to 1998

Abstract

Annual counts of nests with eggs or chicks (known nests) were made at blue penguin (Eudyptula minor) breeding sites on the Otago Peninsula in each November from 1994 to 1997. Although the population has doubled to an estimated 600 known nests over this period, the number of breeding sites on the Otago Peninsula has reduced since the 1970s. Breeding success at three areas at Taiaroa Head were monitored by regular nest checks in the breeding season from 1992 to 1998. At Taiaroa Head reproductive success ranged from 41 to 78% at the three sites during the seven‐year study and was generally higher for pairs nesting in nest boxes than for those in burrows. The percentage of breeding pairs that laid a second clutch after fledging at least one chick from their first clutch (double brooded) varied between seasons (0–48%) and was correlated with the date of the onset of breeding. Egg loss, possibly through predation by Norway rats (Rattus norvegicus), influenced the significantly lower reproductive success at one area (Area A) at Taiaroa Head during the 1996 season.     

Mortality of Black-tailed Godwit Limosa limosa and Northern Lapwing Vanellus vanellus chicks in wet grasslands: influence of predation and agriculture

Grassland-breeding shorebirds show widespread declines due to a reduction in breeding productivity following agricultural intensification. However, there is also concern that increasing predation causes further declines or precludes population recovery. Predation may itself be enhanced by agriculture through changes in habitat or food availability, but little is known about the mortality of nidifugous shorebird chicks. We studied mortality by radio-tagging 662 chicks of Black-tailed Godwit Limosa limosa and Northern Lapwing Vanellus vanellus in 15 farmland sites in the Netherlands. Tagging and handling had no effect on the condition and survival of godwit chicks, but body condition was reduced by 6–11% in lapwing chicks wearing a tag for longer than 3 days. Fledging success was 0–24% in both species. Mortality was highest in young chicks but remained considerable until after fledging. Losses were traced mostly to predators (70–85%; 15 species, predominantly birds), but at least 5–10% were due to mowing, and 10–20% were due to other causes, including entrapment in ditches and starvation. Chicks staying in fields that were cut before the next radio check were found much more often as mowing victims and somewhat more often as prey remains than chicks in fields not cut, indicating that predation includes a limited amount of scavenging. The predation hazard for godwit chicks was higher in recently cut or grazed fields than in the tall, uncut grasslands they preferred, while that for lapwing chicks was lowest in grazed fields. In godwit chicks, poor body condition increased mortality risk, not only from starvation but also from other causes. Predation on godwit chicks was thus enhanced by intensive farming through a decline in the availability of cover, augmented by a reduced body condition, possibly due to food availability problems. Changes in farming practice may therefore help reduce predation pressure, though the observed interactions explained only part of the high predation rate in godwits and none in lapwings. Predator abundance has increased in Dutch wet grassland regions, and chick predation has become a factor that should be considered in planning the type and location of conservation measures.     

Corvids congregate to breeding colonies of a burrow‐nesting seabird

Egg predation is a major cause of reproductive failure among birds, and can compromise the viability of affected populations. Some egg predators aggregate near colonially breeding birds to exploit the seasonal increase of prey resources. We investigated spatial and temporal variations in the abundance of an egg predator (little raven Corvus mellori; Corvidae) to identify whether ravens aggregate spatially or temporally to coincide with any of three potential prey species: burrow‐nesting little penguin (Eudyptula minor; Spheniscidae), short‐tailed shearwater (Ardenna tenuirostris; Procellariidae), and surface‐nesting silver gull (Chroicocephalus novaehollandiae; Laridae). We derived spatially explicit density estimates of little ravens using distance sampling along line transects throughout a calendar year, which encompassed little penguin, short‐tailed shearwater and silver gull breeding and non‐breeding seasons. High raven abundance coincided temporally with penguin and gull egg laying periods but not with that of shearwaters. The spatial distribution of raven density corresponded with the little penguin colony but not with shearwater or gull colonies. Thus, the presence of little penguin eggs in burrows correlated strongly with little raven activity, and this implies that little ravens may have learnt to exploit the plentiful subsurface food resource of little penguin eggs. Corvid management may be required to maintain the viability of this socially and economically important penguin colony.     

Implications of nest-site limitation on density-dependent nest predation at variable spatial scales in a cavity-nesting bird

Nest‐site limitation may have different implications in the spatial distribution of breeding pairs depending on the availability of suitable habitat and the types of nest‐sites. Distribution of cavities suitable as nest sites may allow circumstantial aggregation or active choice of colonial nesting, which may have different implications on breeding performance through effects on breeding density, with variable costs and benefits depending on the consequences of intraspecific competition, social interactions and predation. We evaluated the effects of breeding density derived from nesting site limitation on breeding performance and predation at different spatial scales and considering multiple social, population and environmental limiting factors in the red‐billed chough Pyrrhocorax pyrrhocorax. The results indicate that variable breeding density may arise within the population depending on the availability and spatial distribution of nest‐sites. Nest‐site availability and distribution may also determine social breeding systems (isolated or aggregated) at variable densities, thus resembling differences found at different spatially distant populations under contrasting environmental conditions. Breeding performance was related to density‐dependent processes of population regulation, especially density‐dependent nest predation due to predator attraction to nest clusters. Results also indicate that predation pressure depend on density patterns at large scales. This suggest that predation may have important consequences on population dynamics of spatially structured populations depending on the strength of this kind of density dependence, which in turn may depend on habitat features affecting the prey but also the spatially variable guild of predators. Because habitat and nesting site availability may vary spatially depending on multiple human influences, understanding the strength and form in which breeding density and nest predation at different spatial scales may influence the size and persistence of populations can help to manage them more adequately.     

Varying response of breeding waders to experimental manipulation of their habitat and predators

Many ground-nesting bird species including waders are of conservation concern across Europe. Population declines of Eurasian Curlew Numenius arquata (hereafter Curlew), a globally Near-threatened wader, exemplify this and are driven by low breeding success due to predation, and loss and degradation of breeding habitat. We tested whether the combined delivery of vegetation management and lethal control of Red Foxes Vulpes vulpes and Carrion Crows Corvus corone and Hooded Crows C. cornix improved Curlew nesting success and breeding abundance across UK study sites over five years. On trial sites, vegetation management improved habitat condition and Curlew distributions shifted towards managed areas. However, changes in predator abundances, Curlew nesting success and Curlew breeding abundance did not differ between trial and reference sites. Northern Lapwing Vanellus vanellus numbers increased on trial sites and decreased on reference sites, whilst Common Snipe Gallinago gallinago numbers showed no treatment response. Some of these outcomes were associated with additional environmental variables added to Treatment:Year terms in three-way interactions (changes in fox abundance associated with abundance of non-native gamebirds; changes in Crow abundance associated with gamebirds and woodland area; changes in Curlew nesting success associated with crow abundance), highlighting that outcomes depend on site-specific contexts, and consistent responses to interventions cannot be assumed. The difficulty in increasing Curlew nesting success likely relates to underlying mesopredator densities in the UK, which are known to be high in a European context. Our results suggest that a model of delivery like this study, including management of predators using lethal control, at comparable intensities, is highly unlikely to be effective for Curlew or Snipe within agri-environment schemes, but could be for Northern Lapwing. To make progress on wader recovery it is imperative to understand the underlying drivers of high mesopredator densities and address these through landscape-scale intervention and policy changes.     

Piping plover chick ecology following landscape-level disturbance

Population declines of disturbance-dependent species due to suppression of natural disturbances are realized across ecosystems. The piping plover (Charadrius melodus; plover), a disturbance-dependent and conservation-reliant shorebird that nests on sandy beaches and barrier islands on the Atlantic Coast, was listed under the United States Endangered Species Act in 1986. In 2012, Hurricane Sandy landed on Fire and Westhampton islands, barrier island nesting sites for plovers in New York, USA. Hurricane Sandy was a natural disturbance in this system, creating abundant nesting habitat. The number of chicks produced by a pair, or a population, is a direct measure of reproductive output, and gaining a better understanding of productivity and chick behavior following large-scale habitat creation may improve plover habitat management and potentially species persistence. We evaluated the effects of landscape features on habitat selection, behavior, and survival of plover broods using logistic regression, generalized linear mixed effects models, and survival models. Plover broods selected flatter sites with less dense vegetation than available at random. Chick foraging rates were highest in moist substrates and were lower in areas of higher nesting plover density. Chick survival was greater for broods that hatched earlier in the breeding season and increased as chicks aged. Generally, providing access to sites with flatter, moist substrates will likely result in higher quality brood rearing habitat on the landscape. Ultimately, vegetation removal and habitat management may be needed to reduce plover nesting density and ensure sufficient habitat, which may in turn improve plover chick survival. Moreover, sustaining natural landscape disturbances such as those resulting from storms, and not taking actions to prevent hurricane-created overwash, will allow these landscape features to persist.     

Impact of predation on king penguin Aptenodytes patagonicus in Crozet Archipelago

Predation can have major effects on population dynamics, but predator–prey interactions in marine ecosystems have rarely been studied. While the king penguin is one of the most studied seabirds, little is known about the impact of predation on its population dynamics. Here, we determine the impact of the main predators (giant petrels and skuas) on king penguin breeding success taking into account the nocturnal predation of petrels. We found that predation is the most important source of breeding failure for king penguins. The smallest chicks within crèches are the most hunted. The periphery of the colony suffers the highest risk of predation during summer. Our study shows the unequal quality of some areas inside the colony in terms of predation risk and breeding success, and points out the importance of timing in successful breeding.     

Growth and decline of a penguin colony and the influence on nesting density and reproductive success

Colonial breeding is characteristic of seabirds but nesting at high density has both advantages and disadvantages and may reduce survival and fecundity. African penguins (Spheniscus demersus) initiated breeding at Robben Island, South Africa in 1983. The breeding population on the island increased in the late 1990s and early 2000s before decreasing rapidly until 2010. Before the number breeding peaked, local nest density in the areas where the colony was initiated plateaued, suggesting that preferred nests sites were mostly occupied, and the area used by breeding birds expanded. However, it did not contract again as the population decreased, so that nesting density varied substantially. Breeding success was related positively to the prey available to the breeding birds and negatively to local nest density, particularly during the chick-rearing period, suggesting a density-dependence operating through social interactions in the colony, possibly exacerbated by poor prey availability when the breeding population was large. Although nest density at Robben Island was not high, nesting burrows, which probably reduce the incidence of aggressive encounters in the colony, are scarce and our results suggest that habitat alteration has modified the strength of density-dependent relationships for African penguins. Gaining a better understanding of how density dependence affects fecundity and population growth rates in colonial breeders is important for informing conservation management of the African penguin and other threatened taxa.     

Habitat selection responses of parents to offspring predation risk: an experimental test

The ability of nest predation to influence habitat settlement decisions in birds is widely debated, despite its importance in limiting fitness. Here, we experimentally manipulated nest predation risk across a landscape and asked the question, do migratory birds assess and respond to variation in nest predation risk when choosing breeding habitats? We examined habitat preference by quantifying the density and settlement date of eight species of migratory passerines breeding in areas with and without intact nest predator communities. We found consistently more individuals nesting in areas with reduced nest predation than in areas with intact predator assemblages, although predation risk had no influence on settlement or breeding phenology. Additionally, those individuals occupying safer nesting habitats exhibited increased singing activity. These findings support a causal relationship between habitat choice and nest predation risk and suggest the importance of nest predation risk in shaping avian community structure and breeding activity.     

Nest-site selection by Common Black-Hawks in southwestern New Mexico

ABSTRACT.   Despite the interest of resource managers and conservationists in the status of Common Black-Hawk ( Buteogallus anthracinus ) populations in the southwestern United States, little is known about their nesting success and habitat requirements. Because such information is essential for effective population and habitat management, I examined the nesting success and nest-site selection of Common Black-Hawks in southwestern New Mexico during 2000 and 2001. Of the 37 nesting attempts in 21 territories, ≥1 young fledged from 25 nests (68%). Comparison of nest-sites and nonused sites suggested that breeding Common Black-Hawks selected nest-sites in areas with a sparser and shorter subcanopy tree layer and in trees with a smaller trunk diameter and a greater minimum crown diameter. These differences appear to be related to variation in forest ages within territories, with nonused sites having fewer, but older, canopy trees than nest-sites. Sites with younger, smaller subcanopy trees may provide forest structure for more effective foraging, whereas the characteristics of younger nest tree canopies may reduce the risk of nest predation or offer more protection from inclement weather. Due to the limited range of this species in the southwestern United States, efforts to encourage the establishment and maturation of riparian forests in Common Black-Hawk breeding areas could be important in sustaining available nesting habitat and, in turn, maintaining or expanding current population levels.     

Nest predation and habitat change interact to influence Siberian jay numbers

We studied Siberian jays, breeding in northern Sweden, to examine the potential for interactions between nest predation and reduced vegetation heterogeneity around nest sites to cause a decrease in jay numbers. Parent behaviour and nests are highly cryptic in the species. Our 12-year data showed, however, that nests had a probability of only 0.46 to be successful and produce at least one nestling. Nest predation was intense and a main cause of nest failure. All predators that could be identified were visually oriented hunters, mostly other corvids able to colonize taiga forest only close to human settlements. Consistent with the idea that predators used visual cues, nest predation increased with parental activity, which suggests that predators used parental provisioning trips to locate nests. Furthermore, a reduction in daily nest survival rates with decreasing amount of nesting cover was more pronounced in areas with high corvid activity as predicted when cover mediates the hunting efficiency of visual oriented predators. Declining temperatures interacted with the effects of habitat characteristics to further reduce daily nest survival rates suggesting that parents were not able to increase nest visitation rates to satisfy the higher energy demands of their nestlings without endangering the nest. Our results identify a mechanism through which predation and human-induced reduction in nesting cover on a larger scale may interact to cause a reduction in Siberian jay numbers larger than expected from habitat loss alone.     

Spatial differences in breeding success in the pied avocet Recurvirostra avosetta: effects of habitat on hatching success and chick survival

I studied the breeding biology of pied avocets Recurvirostra avosetta in natural habitats (alkaline lakes), and in semi-natural sites (dry fishpond, reconstructed wetlands) in Hungary to relate habitat selection patterns to spatial and temporal variation in breeding success. Colonies were initiated earlier in semi-natural sites than in natural habitats, and earlier on islands than on the mainland. Hatching success was higher on islands than on the mainland, in semi-natural sites than in natural habitats, in colonies of at least 15 pairs than in smaller colonies, and for nests initiated earlier than later within a colony. Fledging success was higher in the wet years (1999–2000) than in the dry year (1998), decreased strongly by season in both habitats and increased with average daily temperature in the first week post-hatch in 1999–2000. Most pairs hatching young in semi-natural sites attempted to lead their chicks to feeding areas in natural habitats, whereas no such movement occurred in the opposite direction. Chick mortality due to predation was high during brood movements and only 23% of the pairs moving their young produced fledglings, compared to 43% for pairs remaining in semi-natural sites and 68% for pairs hatching and rearing young in natural habitats (total n=192 broods). These results suggest that semi-natural sites were more suitable for nesting, whereas natural habitats were more suitable for chick-rearing. The opposing trends in habitat-related breeding success between nesting and chick-rearing suggest sub-optimal habitat selection by Pied Avocets due to an incorrect assessment of the potential for successful reproduction of semi-natural sites, which may thus function as ecological traps.     

Habitat edges affect patterns of artificial nest predation along a wetland-meadow boundary

Wetland habitats are among the most endangered ecosystems in the world. However, little is known about factors affecting the nesting success of birds in pristine grass-dominated wetlands. During three breeding periods we conducted an experiment with artificial ground nests to test two basic mechanisms (the matrix and ecotonal effects) that may result in edge effects on nest predation in grass-dominated wetland habitats. Whereas the matrix effect model supposes that predator penetrate from habitat of higher predator density to habitat of lower predator density, thus causing an edge effect in the latter, according to the ecotonal effect model predators preferentially use edge habitats over habitat interiors. In addition, we tested the edge effect in a wetland habitat using artificial shrub nests that simulated the real nests of small open-cup nesting passerines. In our study area, the lowest predation rates on ground nests were found in wetland interiors and were substantially higher along the edges of both wetland and meadow habitat. However, predation was not significantly different between meadow and wetland interiors, indicating that both mechanisms can be responsible for the edge effect in wetland edges. An increased predation rate along wetland edges was also observed for shrub nests, and resembled the predation pattern of real shrub nests in the same study area. Though we are not able to distinguish between the two mechanisms of the edge effect found, our results demonstrate that species nesting in wetland edges bordering arable land may be exposed to higher predation. Therefore, an increase in the size of wetland patches that would lead to a reduced proportion of edge areas might be a suitable management practice to protect wetland bird species in cultural European landscapes.     

Diagnosing the environmental causes of the decline in Grey Partridge Perdix perdix survival in France

We studied Grey Partridge Perdix perdix mortality during breeding to identify the environmental causes of a long‐term decline in adult survival. We radiotagged and monitored daily from mid‐March to mid‐September 1009 females on ten contrasting study sites in 1995‐97. Simultaneously, we recorded habitat features and estimated the abundance of Hen and Marsh Harriers Circus cyaneus and C. aeruginosus Red Fox Vulpes vulpes and mustelids. We experimentally tested whether scavenging could have biased predation rates. We also examined, through the necropsy of 80 carcasses of Grey Partridge, whether disease, parasites or poisoning could have been ultimate causes of high predation rates. The survival rate of radiotagged females during spring and summer ranged from 0.25 to 0.65 across study areas. Mortality peaked in May, June and July when females were laying and incubating. The direct negative impact of farming practices was low (6%). Predation was the main proximate cause of female mortality during breeding (73%) and determined the survival rate, suggesting no compensation by other causes of mortality. Ground carnivores were responsible for 64% of predation cases, and raptors for 29%, but this proportion varied across study sites. Disease and poisoning did not appear to favour predation, and scavenging was not likely to have substantially overestimated predation rates. The predation rate on breeding females was positively correlated with the abundance of Hen and Marsh Harriers, suggesting an additional mortality in areas where harriers were abundant. The proportion of raptor predation was linearly related to harrier abundance. The predation rate was not correlated with the abundance of the Red Fox and mustelids. A potential density‐dependent effect on the predation rate was confounded by the abundance of harriers. We found no convincing relationship between the predation rate and habitat features, but we observed a positive relationship between the abundance of Hen and Marsh Harriers and the mean field size. This suggested that habitat characteristics may contribute to high predation rates through predator abundance or habitat‐dependent predation.     

Effect of predation by skuas on breeding success of the Cape petrel Daption capense at Nelson Island, Antarctica

Breeding success of Cape petrels at Nelson Island (South Shetland Islands) in 1991/1992 averaged 29%. Predation by skuas accounted for 76% of all nesting failures. Direct attacks prevailed over opportunistic predation. Intensity of observer disturbance alone could not account for the inter-colony variation in predation rate, which tended to increase with colony size, presence of nearby breeding skuas and local availability of other prey species (penguins). Predation was significantly lower on nests that could be attacked only, or more easily, from the air and nests with close neighbours. Selective predation on nests in which relatively narrower eggs were laid was independent of nest-site characteristics and laying date, suggesting an effect of parental age/experience on breeding success through antipredator behaviour. Development of oil-spitting behaviour in growing chicks balanced the parental attendance and effort in nest defence, which declined as chick age increased. Increased predation on lighter but not smaller chicks may reflect the lower ability of such chicks to defend themselves by spitting stomach oil. Seasonal variations of predation rate on Cape petrel nests did not result from predator swamping by numbers of simultaneously active nests, but rather reflected temporal availability of alternative prey. Success of all phases of nesting declined with later laying date. Received: 15 December 1997 / Accepted: 11 April 1998     

Comparison of population numbers of yellow-eyed penguins,

During a comprehensive survey in 1999, 2000 and 2001, we investigated the number of breeding yellow-eyed penguin pairs on Stewart Island, where cats are present, and on adjacent cat-free islands. We found 79 pairs of yellow-eyed penguin breeding in 19 discrete locations on Stewart Island (4.2 pairs per location), and 99 pairs breeding in 10 discrete locations on all cat-free islands (9.9 pairs per location). Large-scale humaninduced habitat modifications have not occurred on Stewart Island, nor on any of its adjacent offshore islands. While the extensive coastline of Stewart Island (673 km) offers potentially large areas of breeding habitat for penguins, the highest number of breeding pairs were found on the smaller, predator-free Codfish Island (25 km coastline), where a total of 61 breeding pairs were recorded. On Stewart Island, where mustelids do not occur, only feral cats can pose a serious threat to penguin offspring. Results from this study suggest that feral cats may prey on yellow-eyed penguins on Stewart Island. Further work is necessary to investigate whether the observed low numbers of yellow-eyed penguins on Stewart Island are caused by feral cat predation. If so, it may be possible to develop appropriate measures to protect this penguin species from a population decline.     

The influence of habitat management on the breeding success of the Great Bittern Botaurus stellaris in Britain

Great Bitterns Botaurus stellaris have experienced a population decline in the UK, such that in 1997 the total number of breeding males was just 11. This study aimed to identify factors affecting productivity, and how management could be used to manipulate this. An intensive study of Great Bittern breeding success was conducted between 1997 and 2001. The date that males established their booming territories was closely correlated with when females started nesting. Wetter sites with greater fish densities had males that established their booming territories earlier in the season. However, only the date that males started booming determined when females started nesting. The mean clutch size of Great Bitterns was four and the only cause of nest failure was predation. Of eight nests suspected to be second attempts the mean interval between these first and second attempts was 12.25 ± 0.88 (sd) days (range 8–15). The fate of radiotagged chicks followed to fledging revealed that the overall probability of a chick surviving to fledge was 39.1%. Daily losses of Great Bittern chicks due to starvation/exposure accounted for 76.25% and predation 21.25%. The youngest chicks in poor condition were most likely to die, particularly in periods of high rainfall. A simulated renesting model allowed estimation of Great Bittern productivity as 1.24 chicks per female and 1.52 nesting attempts per female. Habitat management, or lack of water control that resulted in sites being drier in spring, delayed nesting, although statistically there was no difference in productivity compared with wetter sites.     

Nesting failure in Finnish Northern Goshawks Accipiter gentilis: incidence and cause

Habitat composition is an important factor influencing nesting failure probability in birds. However, although various habitat effects such as fragmentation and edge density are known to have clear negative effects on the breeding success of passerines, the role that habitat composition plays in shaping nesting failure patterns among other avian groups is less well known. We studied nesting failure in a large forest raptor, the Northern Goshawk Accipiter gentilis , during the period 1999–2003 in Finland. Illegal human persecution was found to constitute a major cause of failure among nests where cause could be determined accurately (comprising c . 60% of all failed nesting attempts). Egg predation by corvids and nestling predation by Eagle Owls Bubo bubo were also common. However, the exact cause of nesting failure could not be inferred in every case. Overall, the degree of initial parental investment (clutch size and egg volume) was significantly smaller in nests that failed than in nesting attempts that were successful. This did not apply to nests that were destroyed by humans, in which investment was at a level equal to nests that were successful. Although the probability that a nesting attempt would fail was also related to small-scale nest concealment, nesting failure probability was not associated with main prey density or several measures describing territorial habitat composition at larger scales. Small initial parental investment, not habitat composition, is thus the major correlate of nesting failure in Goshawks.     

Safety first: terrestrial predators drive selection of highly specific nesting sites in colonial‐breeding birds

Nesting is a critical yet hazardous life stage for many birds. For colonial‐breeding birds, the conspicuousness of the colony to predators suggests immense pressure to select optimal colonial nesting sites. But what drives selection of those sites? As with solitary nesting birds, reducing access by predators may be the single most important factor. If so, knowledge of the predators involved and the attributes of different potential colony sites can allow us to predict the features that make a site especially safe. We examined the attributes of trees used by breeding colonies of metallic starlings Aplonis metallica in tropical Australia, and experimentally tested if those attributes prevented nest access by predatory snakes. Our surveys confirmed that tree choice by starling nesting colonies is highly non‐random, with all colonies located in tall trees in rainforest clearings, with no low branches and smooth bark. Experimental tests demonstrated that the climbing ability of predatory snakes depends upon bark rugosity, and that colony access by snakes depends on tree attributes such as bark rugosity and canopy connectivity. Our study confirms that colonial‐nesting starlings select colony sites that provide a safe refuge from predation. Intense predation pressure may have driven the evolution of stringent breeding habitat criteria in many other species of colonial‐breeding birds.