Nest predation is an important factor in the ecology of passerines and can be a large source of mortality for birds. I provide an overview of factors affecting nest predation of passerines in North America with the goal that it may provide some insight into the ecology and management of woodland birds in the United Kingdom. Although several factors influence productivity, nest success is perhaps the most widely measured demographic characteristic of open-cup-nesting birds, and nest predation is usually the largest cause of nest failure. The identity of predator species, and how their importance varies with habitat and landscape factors, must be known for managers and scientists to design effective conservation plans and place research on nest predation in the appropriate context. Recent studies using video surveillance have made significant contributions to our understanding of the relative importance of different predator taxa in North America. Spatial and temporal variation in nest predation can be better understood when landscapes are placed in a biogeographical context and local habitat and nest-site effects are placed in a landscape context. Low productivity resulting from high nest predation is one of several potential causes of bird population declines in North America and the UK. Although the ‘forest fragmentation paradigm’ from the eastern US may not apply directly to the UK, thinking about avian demographics from a multiscale perspective, and consideration of factors affecting nest predation with knowledge of the dominant predator species, may provide insight into population declines.
Despite the acknowledged importance of nest predation to bird populations, few studies have directly identified nest predators or determined their relative importance (but see Thompson et al. 1999, Pietz & Granfors 2000). The relative importance of nest predator species and levels of nest predation can be affected by nest-site, habitat patch, landscape and regional or biogeographical factors. Results for individual studies, however, have found varying support for these factors, and sometimes have found effects operating in different directions. Thompson et al. (2002) proposed a hierarchical, multiscale model to explain observed spatial and temporal variability in nest predation. They hypothesized that larger scale factors (biogeographical or landscape) provide context for, and may constrain, effects at local or nest-site scales. Observations that larger scale factors (landscape) are stronger than those related to habitat patch, edge or a nest-site provides some support for this model (Donovan et al. 1997, Chalfoun et al. 2002, Thompson et al. 2002, Stephens et al. 2003).
Here, I review factors affecting predation of songbird nests. My review focuses on open-cup-nesting passerines in North America because of my experience with these birds and because the bulk of the literature addresses these species. I review literature on the importance of nest predation and predators to songbirds and on the effects of factors at multiple spatial scales on levels of nest predation and predator abundance and activity. This is not an exhaustive literature review because other more thorough, narrowly focused reviews exist on topics such as the role of predation in edge effects and fragmentation effects (e.g. Robinson & Wilcove 1994, Faaborg et al. 1995, Walters 1998, Thompson et al. 2002). Rather, I provide an overview of ideas and new research in this area in North America that may be relevant to the ecology and management of woodland birds in the United Kingdom.
IMPORTANCE OF NEST PREDATION
Productivity, or fecundity, typically expressed as number of young produced per female per year, is a key vital rate affecting whether a population is increasing or decreasing. In passerines, annual fecundity is a function of clutch size, nest success (the survival of eggs and nestlings), number of nest attempts (following nest failures) and the number of broods (successful nest attempts) (Ricklefs 1973). During the last 20 years nest success has been measured by many studies as an indication of productivity because it is generally easier to monitor the success of nests than the number of young produced by pairs. Long-term demographic studies (Holmes et al. 1992) and population modelling (Schmidt & Whelan 1999, Donovan & Thompson 2001) demonstrate that number of nesting attempts, number of broods and number of fledglings are all important determinants of productivity. Nest predation is the primary factor affecting these components of productivity but, in some systems or in some years, food availability may explain more variation in productivity than nest losses to predation (Nagy & Holmes 2004). Nevertheless nesting success is an important determinant of productivity and perhaps the most widely measured demographic characteristic of open-cup-nesting birds.
In situations where nest predation is identified as important, further investigation of the factors involved inevitably requires the identification of the key predator species. Investigations of factors affecting nest predation should be undertaken or interpreted with knowledge of who the nest predators are; however, for most bird species we have insufficient knowledge of this. The importance of factors affecting nest success varies among studies, and this is probably the result of predator diversity and differences in predator communities (Tewksbury et al. 1998, Marzluff & Restani 1999, Thompson et al. 2002). Investigators should take into consideration the identity of dominant predators and their behaviour when developing hypotheses concerning factors affecting nest predation. In circumstances where mitigation of nest predation is considered necessary, knowledge of predator species and predator-specific management will probably be required.
I reviewed eight studies that used video surveillance to identify predators of eggs and nestlings of open-cup-nesting passerines in temperate North America and compiled the number of depredation events by predator species and the habitats in which they occurred (grassland, shrubland, forest) (Table 1). Mammals, snakes, birds and insects were responsible for 88, 86, 52 and 16 of 242 depredation events, respectively, and at least 40 species of predators were recorded. Even this small number of studies identified a great diversity of predators with relatively even representation of mammals, snakes and birds. However within individual studies, habitat or geographical regions, a particular predator group was often dominant. For example, snakes were the dominant predators in midwestern, southern and western shrubland or scrub habitats (Morrison & Bolger 2002, Stake & Cimprich 2003, Thompson & Burhans 2003); small and medium-sized mammals were the dominant predators in northern grasslands (Pietz & Granfors 2000); and in fragmented forest (Thompson & Burhans 2003) and grasslands (Renfrew & Ribic 2003) a more even mix of snakes, birds and mammals occurred, many of which are associated with edges (Table 1, Fig. 1). Insects (fire ants) only occurred as predators in Texas (Stake & Cimprich 2003, Stake et al. 2004). Thompson and Burhans (2004) was the only study among those above that directly measured and compared the importance of predators in two habitats; they clearly showed the dominance of snakes in shrubland habitat compared with a more even representation of predator taxa in forest habitat (Fig. 2).
Table 1. Number of depredation events of eggs and nestlings of open-cup-nesting passerines in North America by predator species identified by video surveillance of real nests in different habitats
Thompson et al. (2002) proposed a hierarchical, multiscale model to explain spatial and temporal variability in nest predation. They hypothesized that large-scale factors (biogeographical or landscape) provide context for, and may constrain, effects at habitat patch or nest-site scales (Fig. 3). I believe this model greatly facilitates understanding nest predation and explaining the variability among results of previous studies by placing factors that affect predator communities in a biogeographical and landscape context.
Nest-site, habitat and landscape factors affecting predation should be expected to vary as predator communities vary geographically. An example of this is latitudinal patterns in the relative importance of snakes and small mammals as nest predators. I plotted the location of four studies that identified predators of passerine nests in grassland or grassland–shrub habitats in the midwestern United States on a map and listed the most common nest predators in decreasing order of importance (Fig. 1). The importance of snakes and small mammals as nest predators decreased and increased, respectively, with increasing latitude. Despite the small sample size, this pattern is consistent with the expectation that the diversity and abundance of snakes should decrease with latitude. Complicating this gradient, however, is the trend among these study sites for more woody cover in the two most southern sites, and the potential effects of different nesting bird species and alternative prey.
The type and pattern of habitat patches in the landscape can significantly influence the number and diversity of predators, as well as constrain the importance of more local-scale habitat factors such as patch size, vegetation structure or distance to edge effects on nest predation (Thompson et al. 2002). Increased nest predation is often associated with increased fragmentation in forests (Marzluff & Restani 1999, Thompson et al. 2002) and grasslands (Johnson & Temple 1990, Winter & Faaborg 1999). In fact, two independent reviews of existing evidence for fragmentation effects at a landscape scale have concluded that sufficient evidence exists to support a paradigm of negative effects of landscape-scale fragmentation on nesting success (Thompson et al. 2002, Stephens et al. 2003). Thompson et al. (2002) hypothesized the mechanism for forest fragmentation effects is that the increase of non-forest habitats, typically agricultural and urban land uses, or the juxtaposition of non-forest and forest habitats increases the carrying capacity of the landscape for predators.
Robinson et al. (1995) and Donovan et al. (1995) were the first to demonstrate landscape-level effects of forest fragmentation with empirical data from real nests. They related the percentage of forest cover within a 10-km radius (a simple measure of forest fragmentation) of study sites in five midwestern states to daily nest predation. For all nine species examined by Robinson et al. (1995) predation decreased with increasing percentage forest cover; the regression coefficient for percentage forest cover was significant for three species (P < 0.05), two additional species had P-values between 0.05 and 0.20, and a combined probabilities test on all nine species indicated the overall effect of percentage forest cover was significant (P < 0.02). I present graphs for two of the species with significant effects, and two with marginally significant effects (Fig. 4).
In a rigorously designed observational study, Donovan et al. (1997) confirmed predation rates increased with forest fragmentation, and fragmentation (landscape) effects overwhelmed local edge effects. Hartley and Hunter (1998) conducted a meta-analysis of a set of artificial nest experiments and showed that predation rates increased as forest cover decreased at 5-, 10- and 25-km scales of forest cover. Both Donovan et al. (1997) and Hartley and Hunter (1998) addressed factors at multiple scales by investigating the interaction between local edge effects and landscape fragmentation effects, and I discuss this below under edge effects. Donovan et al. (1997) demonstrated that edge effects (discussed below) were dependent on landscape context; in highly fragmented landscapes predation was high throughout the forest whereas in landscapes with low levels of fragmentation predation was high near edges and low in forest interior.
The above studies in eastern North America supporting fragmentation effects were conducted in a similar biogeographical context and hypothesized that predation and parasitism were high in fragmented landscapes as a result of increases in the abundance of generalist predators and Brown-headed Cowbirds Molothrus ater, which were associated with agricultural (Donovan et al. 1995, Robinson et al. 1995, Thompson et al. 2000) or urban habitats (Bakermans & Rodewald 2006). By contrast, Tewksbury et al. (1998) reported that levels of predation at real nests increased with higher landscape levels of forest cover in western North America. They believed the primary predator in their landscape was the Red Squirrel Tamiasciurus hudsonicus, which was more abundant in heavily forested landscapes; I believe this is a good example of how biogeographical differences in predator communities can result in apparently contrary outcomes. More recently, Tewksbury et al. (2006) demonstrated additive effects from the amount of both agricultural and woodland land cover in the landscape on nest predation of western riparian birds, presumably because of combined effects of generalist predators associated with agriculture (similar to the fragmentation effect) and native woodland predators. George and Dobkin (2002) suggested that birds in western North America may be less affected by fragmentation because they have contended with the natural heterogeneity of western landscapes for thousands of generations.
Habitat and nest-site factors
Factors such as habitat type, vegetation density or cover, and distance to edge are characteristics of a habitat patch and factors such as nest height and nest concealment are characteristics of the nest-site. The effects of habitat patch characteristics on nest success, especially distance to edge, vary among species, habitats and regions (reviewed in Paton 1994, Andrén 1995, Lahti 2001, Batary & Baldi 2004); however, the general conclusion to be drawn from these studies is that there is support for edge effects in some landscape or biogeographical contexts. Some studies have found a relationship between nest success and riparian corridor width/patch size (Chapa 1996, Vander Haegan & DeGraaf 1996, Knutson et al. 2000, but see Tewksbury et al. 1998) and between nest success and edge type (Chapa 1996, Suarez et al. 1997, Saracco & Collazo 1999, but see Tewksbury et al. 1998). Peak et al. (2004) and Knutson et al. (2004) did not find strong effects of riparian forest corridor width or patch area, and suggest that patch or width effects will not be detectable in highly fragmented landscapes because nest predation rates are high throughout habitat patches in these landscapes.
Predator communities and predator-specific mortality rates vary among habitat type (discussed previously), so it follows that overall nest predation or mortality can as well. For example, Fink et al. (2006) found species-specific differences in nest success of shrub-nesting birds in young regenerating forest compared with cedar glades, a fire-maintained shrubland habitat. Differences in nest predation among habitats could result from differences in predator communities, bird and alternative prey densities, and vegetation structure and composition.
Temporal factors affecting nest predation can include year, time of season (date) and stage of nest cycle (laying, incubating, nestling) or nest age. I address temporal factors outside the multiscale framework described above because temporal variables are largely a ‘nuisance parameter’ that act as a surrogate for other unmeasured factors. There is no reason to believe time itself is an important factor affecting nest predation other than as time progresses, the cumulative time a nest is exposed to predation increases. Temporal effects on nest predation are detected more often than nest-site factors (Nolan 1963, Roseberry & Klimstra 1970, Best 1978, Zimmerman 1984, Vickery et al. 1992, Budnik et al. 2002, Burhans et al. 2002, Peak et al. 2004, Grant et al. 2005), perhaps because they are indicative of predator or prey community effects that are not more directly addressed.
Year effects on nest predation may account for changes in predator and prey populations or habitat conditions that are ultimately responsible for predation levels. Time of season may account for variation in the numbers of nests available, changes in population levels or behaviour of predators, availability of alternative prey or changes in habitat or nest-site conditions. Nest predation often decreases during the nesting season and is sometimes lower for very early nests as well (Burhans et al. 2002, Grant et al. 2005).
Time within the nest cycle can be related to nest stage and variation in cues provided by prey (eggs or nestlings) or parents. Eggs may be more exposed during laying than incubation, nestlings may provide more cues than eggs and older nestlings may provide more cues than younger nestlings. Parental behaviour, such as the amount of time spent on the nest or number of trips to the nest, may vary the cues provided to predators. Several studies have found a decline in nest success during the nestling stage (Young 1963, Robertson 1972, Schaub et al. 1992, Burhans et al. 2002) and deduced that begging nestlings or feeding trips by parents provided increased cues to predators. Other studies, however, found greater predation during laying or incubation (Holcomb 1972, reviewed in Martin 1992, Sockman 1997, Peak et al. 2004). Grant et al. (2005) related nest success to nest age in days; nest success was lowest shortly after hatch and then increased during the nestling stage. They suggested that the sudden visual and auditory cues associated with parental feeding and newly hatched young increased cues to predators and that as nestlings grew they become less susceptible to non-predation losses and greater parental and self-defence reduced losses to predation.
PREDATOR ABUNDANCE AND BEHAVIOUR
Landscape, habitat and temporal effects on nest predation are ultimately the result of numerical or functional responses by predators, although few studies have directly tested these responses. In general, functional (behavioural) predator responses occur at local and patch scales, while numerical (predator abundance) responses occur at the landscape scale (Chalfoun et al. 2002). Only a few studies have directly investigated landscape effects on predator abundance or distribution. Andrén (1992) surveyed corvid numbers and monitored predation of artificial nests across a gradient of forest fragmentation in Sweden; corvid density and predation increased with forest fragmentation. Raccoons Procyon lotor and Opossums Didelphis virginiana reach their highest densities in fragmented landscapes in Missouri (Dijak & Thompson 2000), potentially because their distributions are associated with developed and agricultural habitats that are interspersed with forest habitat. Rosenberg et al. (1999) surveyed occurrence of some potential nest predators along with tanager species; they generally found positive relationships between predators and fragmentation but responses were often region- or species-specific.
A few studies have investigated the distribution of nest predators in relation to habitat patch characteristics such as use of habitat edges. For example, King et al. (1998) determined that the distribution of Eastern Chipmunk Tamias striatus and Red Squirrels was affected by forest composition and edge, whilst in Missouri Raccoons are detected more frequently near riparian and agricultural edges than in forest interior (Dijak & Thompson 2000). Several studies suggest snakes also prefer habitat edges (Weatherhead & Charland 1985, Durner & Gates 1993, Chalfoun et al. 2002). Three avian predators in Missouri, Brown-headed Cowbird, Blue-jay Cyanocitta cristata and American Crow Corvus brachyrynchos, are also detected more frequently in the forest edge than in the interior (Chalfoun et al. 2002).
Chalfoun et al. (2002) conducted a meta-analysis of landscape, patch and edge effects on the abundance or activity of potential nest predators. Effects were more prevalent among studies at the landscape scale than patch scale, and also at the patch scale compared with the edge scale. This spatial hierarchy of effects nicely parallels the multiscale model for nest predation effects proposed by Thompson et al. (2002).
Nest predation is a major factor affecting bird communities. For example, in North America, nest predation is a key conservation issue for woodland birds nesting in fragmented midwestern forests (Robinson et al. 1995). The identity of predator species, and how their importance varies with habitat and landscape factors, must be known for managers and scientists to design effective conservation plans and place research on nest predation in the appropriate context (Marzluff et al. 2000, Heske et al. 2001, Thompson & Burhans 2003). Spatial and temporal variation in nest predation can be better understood when landscapes are placed in a biogeographical context and local habitat and nest-site effects are placed in a landscape context. For forest songbirds in the eastern United States this multiscale model and the ‘forest fragmentation paradigm’ provide useful guidance for understanding population dynamics and for implementing conservation.
Several important differences and similarities exist in land cover and bird communities between North America and the UK that affect the relevance of these conclusions to the UK. Woodlands once covered most of the UK but 5000–7000 years of human use reduced them to approximately 8% of the land area by 1950 (Peterken 1981). The remaining woodland was highly managed to provide valuable products for the growing human population (Rackham 2006). Although there are no known examples, it is likely that species highly sensitive to forest area or fragmentation were probably extirpated many years ago or adapted to these new landscapes. In recent years forest cover has increased slightly to 12% of the land area. As in North America, there are concerns for declining woodland bird populations. Ten out of 32 species of woodland birds declined by more than 50% between 1966 and 1999 (Fuller et al. 2005) and a re-survey of woodlands in 2003–04 indicated eight of 34 woodland species had declined more than 25% (Amar et al. 2006). There is no single over-arching hypothesis for declines but potential factors include: losses during migration or winter, climate change on breeding grounds, reduction in invertebrate foods, impacts of land use on woodlands, reduced management of lowland woodlands, deer browsing, increased predation pressure from predators such as Grey Squirrel Sciurus carolinensis (Fuller et al. 2005, Amar et al. 2006).
In contrast, in the United States most forest cover was lost much more recently (in the 1800s and early 1900s). Forest landcover has remained fairly stable (approximately 30% of the land area) for the last 25 years, and some regions are much more forested than this (e.g. 76% in the northeastern United States). As in the UK, there is no conclusive evidence as to what factors are responsible for declines in North American forest songbirds. Similar to the UK, changes in habitats due to succession may be responsible for declines in some species, especially early-successional or shrubland species (Askins 2000, Brawn et al. 2001, Hunter et al. 2001), and perhaps some mid-successional forest species. However, for other declining woodland species such as the Wood Thrush Hylocichla mustelina there seems to be abundant potential habitat. Studies of nesting success indicate that in some local populations productivity is too low to maintain local populations, thus creating sink populations that are dependent on other population sources for immigrants. It is not known, however, whether productivity is so low, or sinks are pervasive enough, that low nesting success is driving global declines in any North American species. This remains an important knowledge gap for effective conservation efforts in North America.
I suggest that investigations into declines in woodland birds in the UK should include research and monitoring of species demographics (productivity and survival) in addition to monitoring abundance and habitat use. Although there are many investigations into nesting success and factors affecting nest predation of open ground species in the UK, such studies of woodland birds appear less common than in North America (especially for species that do not nest in nestboxes or cavities). Although the ‘forest fragmentation paradigm’ from the eastern United States may not apply directly to the UK, thinking about avian demographics from a multiscale perspective, and considering factors affecting nest predation with knowledge of who the dominant predators are may provide some useful insight for conservation.
I thank my many colleagues and graduate students who contributed to this research. I thank Rob Fuller and Ken Smith for their invitation to present this work and to contribute to the proceedings, and for their comments on the manuscript. I thank Wes Bailey and Dirk Burhans for their contributions to the manuscript.