A review of the threats to adult survival for swallows (Family: Hirundinidae)

ABSTRACT

Capsule: For declining migratory birds, including many aerial insectivores, such as swallows, there is evidence that adult survival is a demographic process with strong effects on population trends.

Aims: The aim was to identify and quantify the effect of threats affecting adult survival and potentially driving population declines for five well-studied swallow species: Barn Swallow Hirundo rustica, Cliff Swallow Petrochelidon pyrrhonota, Tree Swallow Tachycineta bicolor, Sand Martin Riparia riparia, and Purple Martin Progne subis.

Methods: We reviewed the literature to identify the threats to adult survival, quantified the magnitude of the effect and identified whether threats had a direct or indirect effect on survival.

Results: We identified habitat change, weather, competition, incidental loss, contaminants, insect availability, disease, and predation as threats to adult survival in swallows, although for many of these threats there was limited information to quantify their impact. However, weather, particularly cold snaps and precipitation, had negative effects on survival for many populations of four species, either directly or indirectly through effects on insect availability. When there was a relationship, weather was associated with a 13–53% decrease in survival.

Conclusion: Based on the available research, weather conditions throughout the annual cycle is a key threat to adult survival for several swallow species. However, future research on the threats to these species should consider examining the effect of insect availability and the effect of threats during the non-breeding period on survival. Finally, we suggest that new research should be devoted to understanding the importance of adult survival for declining bird populations.

Population declines across many species are often driven by a diversity of threats (Dulvy et al. 2014, Purcell et al. 2014, Paleczny et al. 2015). Identifying and addressing the threats with the greatest effect on each species is challenging. One approach to understanding the effect of threats on populations is to examine the relationships between threats and demographic processes, such as fecundity, recruitment, and survival, which provide insights into how threats drive declines (Rappole & McDonald 1994, Rushing et al. 2016, Selwood et al. 2015). In particular, by focusing on the threats to demographic processes that have the greatest effect on population dynamics, it is possible to direct conservation efforts toward mitigating the threats with the greatest effect on populations (Green 1999).

Life history theory predicts that population trends in species with a short life span and high reproductive output should be driven by fecundity (Wilson & MacArthur 1967, Sæther & Bakke 2000). This prediction has been upheld for some migratory passerines (Ambrosini et al. 2011, Gardali et al. 2000), a group of birds that typifies this pace of life. However, for other species of migratory passerines, particularly those experiencing population declines, population dynamics are driven by adult survival (Baillie & Peach 1992, Murphy 2001, Fletcher et al. 2006, Buehler et al. 2008, Perlut et al. 2008, Ambrosini et al. 2011). Specifically, for these species, decreased adult survival is the demographic process driving population declines. For example, using a population projection model with sensitivity and elasticity analyses, population trends for Cerulean Warblers Dendroica cerulea were found to be most strongly related to changes in adult survival, rather than breeding success or juvenile survival (Buehler et al. 2008). Additionally, this model showed that conservation efforts aimed at increasing breeding success to offset reductions in adult survival could not fully compensate for the loss of adults in the population (Buehler et al. 2008). Despite the importance of adult survival on population trends, research on declining bird populations is often focussed on the effects of threats and other regulating factors (collectively referred to as threats hereafter) on fecundity.

One group of birds presently experiencing steep declines in parts of North America and Europe are aerial insectivores (Inger et al. 2015, Michel et al. 2016, Nebel et al. 2010, Smith et al. 2015). While the cause(s) of the declines are still unknown, considerable research has examined the effects of threats on fecundity in swallows (Fassbinder-Orth et al. 2013, Møller 2013, Winkler et al. 2013, Rioux Paquette et al. 2014). Indeed, a Biological Abstracts (Thomson Reuters) search for published papers on fecundity for swallows and martins returned 1840 results, whereas a similar search for adult survival returned 132 results. While not all of those results may be directly pertinent to the role of threats on population declines, it does suggest a bias in the types of studies that are conducted. More importantly, it affects the inferences we can make about the role of different demographic rates on population dynamics.

Despite the lack of research on adult survival, several recent papers suggest that local population declines in some swallow populations may be driven by reductions in survival rates, particularly for adults (Cox et al. 2018, Imlay et al. 2018b, Norman & Peach 2013, Schaub et al. 2015, Taylor et al. 2018). Four of these studies examined the vital rates that drive population dynamics (often with an integrated population model or demographic model) in Palearctic-African populations of Barn Swallow Hirundo rustica and Sand Martin Riparia riparia, and two Nearctic-Neotropical populations of Tree Swallow Tachycineta bicolor (Cox et al. 2018, Norman & Peach 2013, Schaub et al. 2015, Taylor et al. 2018). Their results demonstrate that breeding population size was strongly related to adult survival, with large populations when survival was higher. In some of these studies, population dynamics were also influenced by juvenile recruitment (Cox et al. 2018) or immigration (Schaub et al. 2015, Taylor et al. 2018). However, Taylor et al. (2018) noted that the effects of immigration on population size may also be due to lower adult survival rates in source populations which reduce the availability of potential immigrants. Finally, reduced breeding success in recent years has only been observed for one of four Nearctic-Neotropical swallow species, the Sand Martin, whereas breeding success was either unchanged or higher for the other three species (Imlay et al. 2018b). This suggests that population declines for at least three species must be due to reductions in other demographic rates (Imlay et al. 2018b). While the most influential demographic rate for population trends can vary across a species range (Weegman et al. 2017), collectively, the findings from the above studies, along with the large spatial scales covered by this work, provide strong support for the importance of understanding and addressing the threats to adult survival in swallow populations.

Therefore, our broad goal was to identify threats to adult survival throughout the annual cycle for five species of swallows: Barn Swallow, Cliff Swallow Petrochelidon pyrrhonota, Tree Swallow, Sand Martin, and Purple Martin Progne subis. These species are all declining in at least part of their geographic range (Figure 1) and adult survival could be a key demographic process in the population declines (Cox et al. 2018, Imlay et al. 2018b, Norman & Peach 2013, Schaub et al. 2015, Taylor et al. 2018). We conduct a literature review and compare threats based on magnitude of effects on adult survival and determine whether specific threats have direct or indirect effects on survival. We used a multi-species approach to determine if there was evidence of common threats among species and to identify knowledge gaps that, if addressed, may help to understand and address population declines.

Figure 1. Annual population trends using the hierarchical models described by Link and Sauer (2002) with the 95% credible intervals from the North American Breeding Bird Survey for Barn Swallow, Cliff Swallow, Tree Swallow, Sand Martin, and Purple Martin populations from 1966 to 2015. The annual population trend is represented with a log scale and data is from Sauer et al. (2017).

Methods

Focal species

Our five focal species, Barn Swallow, Cliff Swallow, Tree Swallow, Sand Martin, and Purple Martin, are all New World passerines; the Barn Swallow and Sand Martin also have Old World populations. All five species typically breed in their second year and live for a maximum of 8–13 years, depending on the species and population (Brown & Bomberger Brown 1999, Brown et al. 2017, Brown & Tarof 2013, Garrison 1999, Winkler et al. 2011). While most swallows raise one brood each year, Barn Swallows and Sand Martins often raise two successful broods (Brown & Bomberger Brown 1999, Cowley 1983). Throughout their annual cycle, these swallows are often associated with human infrastructure and working landscapes for breeding, foraging, and roosting (Brown & Bomberger Brown 1999, Brown et al. 2017, Brown & Tarof 2013, Garrison 1999, Winkler et al. 2011). Apart from the consumption of some berries (e.g. bayberry and myrtle Morella spp.) by Tree Swallows (Piland & Winkler 2015), swallows solely forage on aerial insects throughout the annual cycle.

Review of threats to adult survival

To determine the documented threats to adult survival in swallows, we searched Biological Abstracts (Thomson Reuters) using keywords associated with the common and scientific names of the focal species (Barn Swallow, Cliff Swallow, Tree Swallow, Sand Martin, Bank Swallow, Purple Martin, Hirundo rustica, Petrochelidon pyrrhonota, Tachycineta bicolor, Riparia riparia, and Progne subis), adult survival (survival, return rate, mortal*, fatal*, lethal*, and impact), and potential threats (habitat, loss, degradation, weather, climate, temperature, precipitation, wind, compet*, density-depend*, road, collision, aggregat*, quarr*, pit, pesticide*, pollut*, insect, food, diseas*, parasit*, predat*, harvest*, and human). The search included studies conducted throughout the species’ geographic range, including New and Old World populations, from 1990 to 2018. We selected this time period because all species experienced population declines across at least part of their geographic range during this time (Heldbjerg & Fox 2008, Sauer et al. 2017) (Figure 1) and because most studies on adult survival for these species were conducted during this time. The literature search was completed on 30 October 2018.

Calculating maximum effects of threat on adult survival

To determine the threats with the greatest effect on adult survival, we calculated the maximum effect of each threat. Given the differences in how effects on adult survival were reported in the literature, we used three different approaches to assess impacts on survival. The first approach used the per cent of the population negatively affected by the threat (sometimes accounting for detectability, e.g. Zimmerling et al. 2013), such as the proportion of the population that died (8 of 25 studies or 32%). In these cases, we expressed the maximum effect as the per cent of carcasses compared to the total population size. The second approach used differences in recapture rates determined with a (generalized) linear (mixed) model or Chi-square test (4 studies or 16%). The third approach used differences in annual or daily survival rates determined with a capture–mark–recapture model (13 studies or 52%). For these latter two approaches, we expressed the maximum effect as the difference in survival between highest (when the threat had no or little effect) and lowest values (when the threat had the greatest effect) for each study.

For all approaches, if the paper provided differences in survival for different groups (e.g. sexes or populations), then we provided a range of values for the maximum effect of the threat on survival. Also, if reported values showed no or negligible effects on survival (defined by authors without data provided) we indicate a value of 0%. Finally, when results were presented as figures, we used ImageJ 1.52a (Schneider et al. 2012) with the Figure Calibration plugin (http://www.astro.physik.uni-goettingen.de/~hessman/ImageJ/Figure_Calibration/ [accessed 19 October 2018]) to obtain values that were then used in our calculations above.

Within each section of the Results below, we provide a summary of the research on each threat, quantify the maximum effect of the threat, and discuss whether the threats have direct and/or indirect effects. Also, where possible, we provide an indication about whether effects of these threats may change in the future.

Results

Our literature review revealed eight broad categories of threats (Table 1).

Table 1. Research documenting the maximum effects of different threats on adult survival or population for five swallow species.

Threat	Species	Measured threat	Effect on survival (%)^a	Stage of annual cycle^b	Method	Reference
Habitat change	Tree Swallow	Pond abundance	12.0	Breeding	CMRa	Clark et al. (2018)
Weather	Cliff Swallow	Cold snap	−53 to 0	Pre-breeding	POP	Brown & Bomberger Brown (1996, 1998)
	Tree Swallow	Cold snap	−17.2	Pre-breeding	POP	Hess et al. (2008)
	Tree Swallow	Cold snap	−25 to −19	Pre-breeding	CMRa	Custer et al. (2012)
	Barn Swallow	Cold snap	−40	Autumn migration	POP	Møller (2011)
	Barn Swallow	ENSO, NAO	−29.4 to 0	Breeding, wintering	CMRa	García-Pérez et al. (2014)
	Tree Swallow	ENSO	0	breeding	CMRa	Weegman et al. (2017)
	Tree Swallow	ENSO	–^c	Wintering	CMRa	Clark et al. (2018)
	Tree Swallow	NAO	−13.0 to 0	Wintering	CMRa	Clark et al. (2018)
	Purple Martin	ENSO	0	Wintering	CMRa	Stutchbury et al. (2009)
	Tree Swallow	Precipitation	−15.0	Pre-breeding	CMRa	Clark et al. (2018)
	Barn Swallow	Rainfall	0	Breeding	CMRa	Robinson et al. (2008)
	Tree Swallow	Temperature, -rainfall^d	−25 to 0	Breeding	CMRa	Weegman et al. (2017)
	Sand Martin	Temperature, rainfall	0	Breeding	CMRa	Norman & Peach (2013)
	Sand Martin	Rainfall	−18	Breeding	CMRa	Cowley & Siriwardena (2005)
	Sand Martin	Rainfall	0	Breeding	CMRa	Robinson et al. (2008)
	Barn Swallow	Rainfall	20	Wintering	CMRa	Robinson et al. (2008)
	Sand Martin	Rainfall	13	Wintering	CMRa	Cowley & Siriwardena (2005)
	Sand Martin	Rainfall	23	Wintering	CMRa	Robinson et al. (2008)
	Sand Martin	Rainfall	13–18	Wintering	CMRa	Szép (1995a, 1995b)
	Sand Martin	Rainfall	43	Wintering	CMRa	Norman & Peach (2013)
Competition	Barn Swallow	Intraspecific	−20	Breeding	RR	Balbontín & Møller (2015)
	Barn Swallow	Intraspecific	0	Breeding	CMRa	Paradis et al. (2002)
	Sand Martin	Intraspecific	−14	Breeding	CMRa	Norman & Peach (2013)
Incidental loss	Cliff Swallow	Vehicles	<−0.1	Breeding	POP	Brown & Bomberger Brown (2013)
	Tree Swallow	Wind turbines	<−0.1	Breeding	POPd	Zimmerling et al. (2013)
Contaminants	Tree Swallow	Mercury	−1.1 to −0.9	Breeding	CMRa	Hallinger et al. (2011)
	Tree Swallow	PCBs	−3.9 to 0	Breeding	CMRa	Custer et al. (2007, 2012)
Disease	Barn Swallow	Feather mites	0	Breeding	RR	Pap et al. (2005)
	Cliff Swallow	Feather mites	−14 to 100	Breeding	CMRd	Brown et al. (2006)
	Purple Martin	Blood parasite	0^e	Breeding	RR	Davidar & Morton (1993)
	Purple Martin	Blood parasites	−22.8 to −5.0	Breeding	RR	Davidar & Morton (2006)
	Purple Martin	West Nile Virus	0	Breeding	CMRa	Stutchbury et al. (2009)
Predation	Barn Swallow	Avian	−0.3	Wintering	POP	Bijlsma & van den Brink (2005)
	Barn Swallow	Feline	−9 to −4	Breeding	RR	Balbontín & Møller (2015)
	Barn Swallow	Human	−20	Wintering	POP	Ash (1995)
	Sand Martin	Avian	−58	Breeding	POP	Young (1994)

Note: Methods used to quantify changes in adult survival included: (1) the percent of the population negatively affected by threat, i.e. the percent of population that died (POP or POPd, when detectability was incorporated into estimate); (2) differences in recapture rates determined with a (generalized) linear (mixed) model or Chi-square test (RR); and (3) differences in annual or daily survival rates determined with a capture–mark–recapture model (CMRa or CMRd, respectively).

^aPositive values reflect a positive relationship between adult survival and increasing values of the measured threat, and negative values reflect a negative relationships between adult survival and increasing values of the measured threat. A range of values is given if survival was reported for different sexes or populations.

^bStage of the annual cycle when the threat was measured; some threats, like contaminants and disease, may have persisted in the individual beyond this stage.

^cNo survival estimates were provided in the text or figures, therefore we are unable to indicate the effect of this threat on survival from this study. However, the results state that a large proportion of the variation in survival could be explained by the relationships with ENSO.

^dIn this study, a weather index of temperature and negative precipitation was used to model survival; with higher values representing warm, dry years.

^eNo values for return rates were provided; however, discussion states chronic infection has little effect on survival, but suggests that mortality due to the initial infection may be high.

Habitat change

We were only able to find one study relating habitat change to adult survival. Lower pond abundance during the breeding season was associated with a 12% decrease in adult survival for Tree Swallows (Figure 2 and Table 1) and was attributed to lower insect availability when ponds were less abundant (Clark et al. 2018). In general, while familiarity with habitats can confer higher survival for swallows (Brown et al. 2008), ultimately, habitat changes have indirect effects on adult survival.

Figure 2. The effects of seven different threats on adult survival for Barn Swallow, Cliff Swallow, Tree Swallow, Sand Martin, and Purple Martin. The effect of each threat on adult survival was measured as the proportion of the population affected, or the difference in highest and lowest return or survival rates. For species where we could only determine the effect on adult survival from one or two studies or subgroups within a study (e.g. sexes, populations), single data points represent the results. Otherwise the effect of each threat for each species is represented as a violin plot with the sample size indicated (i.e. number of separate survival estimates across all studies).

Weather

Many studies have examined relationships between adult survival in swallows and weather throughout the annual cycle, especially in Old World populations. Extended periods of cold temperatures and heavy rain or snowfall during the pre-breeding and migratory periods are associated with mass mortality events for Barn, Cliff, and Tree Swallows (Brown & Bomberger Brown 1998, Hess et al. 2008, Møller 2011), which may directly or indirectly kill as many as 53% of adults in the population (Figure 2 and Table 1). Precipitation in the pre-breeding period is also associated with up to a 15% decrease in survival for Tree Swallows (Clark et al. 2018), but hot-dry conditions during breeding can decrease survival up to 25% (Weegman et al. 2017). In contrast, higher levels of precipitation during breeding are associated with lower survival (up to 18%) for some populations of Sand Martins (Cowley & Siriwardena 2005, but see Norman & Peach 2013), but not Barn Swallows (Robinson et al. 2008). Higher levels of winter rainfall are related to increased survival for Barn Swallows and Sand Martins (Cowley & Siriwardena 2005, Norman & Peach 2013, Robinson et al. 2008). In general, when rainfall has an effect on survival, the maximum effect is a decrease in survival of 13–43% (Table 1).

Conditions associated with large-scale climate indices have mixed effects on adult survival. Two commonly studied indices, the El Niño Southern Oscillation (ENSO) and North Atlantic Oscillation (NAO), measure climatic conditions along a warm/dry – cool/wet or warm/wet – cool/dry continuum, respectively. While there was no relationship between ENSO and Purple Martin survival in one region (Stutchbury et al. 2009), Barn and Tree Swallow survival is often correlated with ENSO and/or NAO with a potential decrease in survival of up to 29% (Balbontín et al. 2009, Clark et al. 2018, García-Pérez et al. 2014). For some populations, cool–dry conditions resulted in lower survival; however there was considerable variation in the relationship between ENSO or NAO and survival across the range of the species in these studies. For example, adult survival for Barn Swallows was related to ENSO and NAO in one population, but not in another population over 3000 km away (García-Pérez et al. 2014). The variability in these relationships may be due to differences in the effect of these indices on weather at different breeding and wintering locations or during migration (Balbontín et al. 2009, García-Pérez et al. 2014) and/or low migratory connectivity between breeding and wintering populations (e.g. breeding populations wintering in a broad geographic area) (García-Pérez et al. 2014, Stutchbury et al. 2009).

Since weather conditions can have large effects on adult survival (13–53%), it is possible that it is contributing to population declines. These effects may be direct and/or indirect as a result of both local and large-scale climatic conditions, for example, the direct effects due to mortality from exposure (Brown & Bomberger Brown 1998, Hess et al. 2008, Møller 2011) and the indirect effects often due to changes in insect availability (Clark et al. 2018, García-Pérez et al. 2014, Norman & Peach 2013). Clearly there is substantial species and geographic variability in how weather affects populations.

Competition

There is limited research on the effect of intraspecific competition on adult survival for swallows and none on interspecific competition. Density dependence can result in up to a 20% decrease in adult survival (Figure 2 and Table 1) for some Barn Swallow and Sand Martin populations (Norman & Peach 2013, Balbontín & Møller 2015, but see Paradis et al. 2002). These density-dependent effects are likely driven by competition between conspecifics for food (Norman & Peach 2013), suggesting that this is an indirect relationship between competition and adult survival. However, other threats may affect this relationship. For example, in Barn Swallows, when predation pressure was low, adult survival was lower at larger breeding colonies, but when predation pressure was high, survival was higher at larger breeding colonies (Balbontín & Møller 2015). Together, this suggests that density-dependent effects vary not only with population size but could interact with other threats.

Incidental loss

Several studies have examined the effects on incidental loss on swallow populations, although only one study quantifies the effects on adult survival. Swallows are frequently observed during road mortality surveys during the breeding season (Erritzoe et al. 2003, Bishop & Brogan 2013), particularly on roads with tree belts or hedgerows as these areas, presumably, provide good foraging sites (Grüebler et al. 2008, Orlowski 2005). Unlike predation, road mortality is more likely to affect individuals in good condition (Bujoczek et al. 2011), resulting in the loss of high quality individuals. In Europe, it is estimated that road mortality affects 1 million Barn Swallows annually, of which roughly one-third are adult (Orlowski 2005). In North America, however, selection on Cliff Swallow wing morphology has reduced road mortality (Brown & Bomberger Brown 2013), with less than 0.1% of adult Cliff Swallows in a population being hit by vehicles. There are other potential sources of incidental losses. A wind turbine study estimated that less than 0.01% of adult Tree Swallows were killed (Zimmerling et al. 2013). Collectively, though, the estimated impacts of incidental losses on adult survival or populations (Figure 2 and Table 1) are relatively small.

Contaminants

Numerous studies have documented high levels of organophosphates, organochlorines, polychlorinated biphenyls (PCBs), and metals with potentially lethal or sub-lethal effects in adult swallows (e.g. Baron et al. 1999, Burgess et al. 1999, Mora et al. 2002, 2012, Custer et al. 2007, Hawley et al. 2009). However, only three studies have examined effects of contaminants on adult survival. Two studies compared survival of female Tree Swallows breeding in areas with high levels of PCBs compared to the control areas. One reported a decrease in survival of up to 3.9% for swallows exposed to higher PCB levels (Custer et al. 2007), but there was no difference in the second study, possibly due to the lower environmental levels of PCBs (Custer et al. 2012). In the third study, there was an up to 1.1% decrease in survival for Tree Swallows exposed to high levels of mercury compared to control groups (Hallinger et al. 2011). Together these results suggest that the contaminants studied thus far have a much lower effect than other threats (Table 1), at least at levels recorded in these studies. The transfer of contaminants to swallows may be directly from their environment or indirectly through insect prey (Fernie et al. 2018, Haroune et al. 2015, Mora et al. 2002). These studies suggest that the effects of the examined contaminants on swallow populations are small.

Insect availability

Despite the importance of aerial insects as the only or primary food source, we were unable to find any research directly relating insect availability to adult survival. However, lower adult survival as a result of habitat change or weather was often attributed to those threats reducing insect availability (Norman & Peach 2013, García-Pérez et al. 2014, Weegman et al. 2017, Clark et al. 2018). The effects of habitat change and weather on insect abundance are well studied. For example, habitat changes from low landscape-level homogeneity (often associated with low intensity agricultural land use, e.g. pastures and hayfields separated by hedgerows) to high landscape-level homogeneity (often associated with intensive agricultural practices, e.g. corn and soy production) reduces insect abundance (Benton et al. 2002, Grüebler et al. 2008, Paquette et al. 2013) and alters diet composition (Orłowski & Karg 2013). These intensive agricultural practices often entail increased pesticide use, which affects the abundance of aerial insects (Morrissey et al. 2015, Pisa et al. 2015, Van Dijk et al. 2013) and possibly leads to changes in diet shown in other aerial insectivores (English et al. 2017). Also, neonicotinoids are implicated in the decline of one Barn Swallow population (Hallmann et al. 2014). Additionally, reduced insect availability during breeding occurs during periods of cold temperatures, rainfall, and high winds (Nooker et al. 2005, Møller 2013, Winkler et al. 2013). Since declines in aerial insect populations have been severe in some regions (Hallmann et al. 2017), it is likely that reduced food availability is contributing to population declines for swallows, but the magnitude of these effects is unknown.

Disease

Many studies have documented the presence of various bacteria, viruses, and parasites (i.e. arthropods, flatworms, nematodes, and protozoa) in adult swallows (Brown et al. 2007, Caron et al. 2014, Davidar & Morton 2006, Heneberg et al. 2011, Stenkat et al. 2014, Von Ronn et al. 2015). These infections often had either beneficial or negligible effects on adult survival for swallows. Feather mites increase Cliff Swallow survival during breeding, presumably through the removal of old preening oil and/or removal or competition with other infectious agents on feathers (Brown et al. 2006), but have no effect on Barn Swallows (Pap et al. 2005). Neither the increased prevalence of West Nile Virus nor chronic infection with a protozoan Haemoptroteus progenei has an effect on Purple Martin survival (Davidar & Morton 1993, Stutchbury et al. 2009), however, the initial protozoan infection may reduce survival (Davidar & Morton 1993). Infection with an unidentified filarial nematode is associated with lower survival in Purple Martins, and, although uncommon, most adults with both infections died (Davidar & Morton 2006). Collectively, these findings suggest that disease has a wide range of direct effects on adult survival (Figure 2) and when they are negative, the effects can result in up to a 5–23% decrease in survival, depending on the specific infection (Table 1).

Predation

Many studies have identified a diverse group of predators (Ash 1995, Bijlsma & van den Brink 2005, Møller & Nielsen 1997, O’Brien et al. 2014, Rebecca 2004), that, in some cases, can directly result in large losses (up to 58%, Figure 2 and Table 1) of adult swallows (Young 1994). Mortality from predation may also be female biased (Møller & Nielsen 1997) which further reduces fecundity and contributes to population declines. The effects of predation on adult survival are, however, generally localized to areas where predators learn to exploit an abundant food source (Ash 1995, Rebecca 2004, Young 1994) and are unlikely to represent broader-scale population-level effects, nor contribute in a substantial manner to population declines.

Discussion

Our literature review suggests that the effects of weather conditions on adult survival in swallows are well studied, resulting in the most conclusive information of all the threats considered. Adverse or extreme weather conditions can result in a 13–53% decrease in adult survival for swallows, depending on the specific weather conditions. Of these conditions, cold snaps or precipitation during the spring pre-breeding period and autumn migration (Brown & Bomberger Brown 1996, 1998, Hess et al. 2008, Møller 2011, Clark et al. 2018) and reduced rainfall during the winter (Cowley & Siriwardena 2005, García-Pérez et al. 2014, Norman & Peach 2013, Robinson et al. 2008, Szép 1995a, 1995b) had the most consistent and negative effects on survival. The mechanism behind lower survival for the latter two relationships is likely indirect, with these weather conditions negatively affecting insect abundance, which, in turn, affects survival.

Although there is strong support for the effects of weather on adult survival in swallows, perhaps, in part, due to the relative ease through which this threat can be studied, there are two important caveats to consider in relation to population declines. First, much of this research, especially during the winter, is on Palearctic-African populations, rather than Nearctic-Neotropical populations. Given the differences in precipitation patterns in Africa, compared to the Neotropics, it is possible that the relationships between winter rainfall and adult survival may not be the same for wintering Nearctic-Neotropical swallow populations. Second, although weather patterns do affect adult survival, this threat can only result in population declines if the frequency of these extreme weather events is increasing. Climate change is associated with changing weather patterns, including more extreme weather events (Hayhoe et al. 2007, Djomou et al. 2015, Screen et al. 2015, Ummenhofer & Meehl 2017), however, no studies have examined long-term data on adult survival for swallows and population trends in relation to the frequency of severe weather conditions. Therefore, it is unknown if declining population trends, in relation to adult survival, are driven by changes in weather patterns.

We also observed that for some Barn Swallow, Sand Martin, and Purple Martin populations, competition, disease, and predation can have large negative effects on survival, e.g. maximum 20%, 22.8%, and 58%, respectively (Table 1). These large effects are often localized to different populations and/or under specific environmental conditions. Furthermore, determining whether these effects are driven by natural or anthropogenic sources is difficult. Therefore, it is important to be careful not to generalize these results across a broader region. Despite these limitations in our findings, we consider that competition, disease, and predation, especially if these factors are increasing with habitat and climate change, could have additive effects on other threats and could be important contributors to lower survival in some regions.

Clearly our findings are limited by several factors. First, we lacked information for some threats, and for different species, particularly throughout their geographic ranges. Second, the approach to measuring survival and threats varied across studies. Survival was measured in terms of the per cent of the population found dead, and in differences in recapture or apparent survival rates. The former two approaches rarely account for detectability, and the latter two cannot distinguish between true survival and permanent emigration. Threats were also not measured on a consistent scale, and many studies, particularly on weather, used different indices to examine weather conditions. Finally, the geographic scale and sample sizes all varied across different studies. Therefore, while our findings provide insight into the known relationships between threats and adult survival, we recognize that these factors limit the scope of our conclusions, particularly for poorly studied threats. However, our review provides important directions for future research to address knowledge gaps and these limitations.

Future research

We were unable to find any papers relating adult survival to insect availability and very little information on the effects of habitat change, incidental loss, and contaminants on adult survival. Three of these threats (i.e. habitat change, contaminants, and insect availability) have received extensive attention in the literature for their effects on fecundity in Barn and Tree Swallows (McCarty & Winkler 1999, Bishop et al. 2000, Ambrosini et al. 2002, Nooker et al. 2005, Fredricks et al. 2012, Winkler et al. 2013, Rioux Paquette et al. 2014, Schifferli et al. 2014, Imlay et al. 2017), but very little is known about their relationships with adult survival. This lack of information is particularly concerning as reductions in insect availability (possibly as a result of habitat change, weather, and/or contaminants) is considered to be the most likely driver of population declines for aerial insectivores (Nebel et al. 2010). Clearly, this represents a key knowledge gap for aerial insectivore declines as a whole. Since demands on food are highest during the breeding season and autumn migration (Kelly et al. 2013), reductions in food availability during these times could have a disproportionate effect on populations and should be investigated.

We also found that for most threats, except weather, there was very little information on their effects during the non-breeding period. This represents another key knowledge gap as non-breeding conditions are strongly related to population trends for several species (Ambrosini et al. 2011, Ockendon et al. 2014, Sicurella et al. 2016). Additionally, for other migratory birds, adult survival rates are lowest during migration (Klaassen et al. 2014, Sillett & Holmes 2002), suggesting that adults may be more vulnerable to threats during this stage of the annual cycle than other stages. The lack of information on threats throughout this period may be partly due to knowledge gaps on migration routes and winter locations for some non-breeding populations of Nearctic-Neotropical swallows. However, this knowledge gap is quickly being addressed with tracking technologies and stable isotopes (Fraser et al. 2012, Hobson et al. 2015, Szép et al. 2017, Imlay et al. 2018a, Knight et al. 2018), and, in the future, it may be possible to estimate survival during different periods of the annual cycle. With an improved understanding of these movements, threats during these times can be identified and related to survival.

Conclusion

Recent research shows that adult survival is a key factor in population trends for some populations of swallows. In our literature review, we identified many threats to adult survival for several declining swallow species and compared the effects of these threats. This review highlights that the effects of threats may across the geographic range of a species and among species. Therefore, caution is clearly warranted when trying to relate environmental conditions during one stage of the annual cycle for a population to broad-scale population declines across a much larger geographic area. Especially as Michel et al. (2016) suggests that populations are affected by multiple threats that vary throughout their range. However, our findings do suggest that at least one well-studied threat, adverse weather, often has similar effects across most swallow populations and species.

As a result of human activities, the effects of at least seven threats (all except competition) are predicted to increase in both their occurrence and severity in the future (Bommarco et al. 2013, Farmer et al. 2008, Hansen et al. 2006, Martinez & Merino 2011, Tilman et al. 2001, Dulac 2013). As threats increase over time and space, they will have a greater effect on adult survival. Also, because many threats are inter-related, their effects on survival are likely to be synergistic rather than additive. For example, adverse weather conditions due to climate change may lead to further habitat loss (beyond that as a result of other human activities) (Mantyka-Pringle et al. 2012) and increased contaminent and disease loads (Hallinger & Cristol 2011, Møller 2011). These synergistic effects could result in an even greater effect on adult survival rates than if the effects were additive, possibly leading to lower adult survival rates in the future.

Developing effective conservation strategies requires understanding when and where species experience threats during the annual cycle and determining how these threats affect demographic processes. However, most research on threats focuses on the breeding grounds (Faaborg et al. 2010, Marra et al. 2015) and on fecundity rather than adult survival. In doing so, conservation efforts may focus on addressing threats that only affect populations during one stage of the annual cycle. Furthermore, for migratory birds, this could result in conservation efforts that do not target the demographic processes that have the greatest effect on population trends.

Acknowledgements

We thank K. Hobson, P. Marra, J. Nocera, L. Philmore, R. Ronconi, and two anonymous reviewers for feedback on an earlier draft of this manuscript.

Additional information

Funding

This work was supported by Natural Sciences and Engineering Research Council of Canada.