INTRODUCTION
Aves, a class represented by around 10,000 species, display a broad diversity of morphologies and behaviors, and also show considerable variation in their lifespan and annual survival. For example, in large-bodied landbirds, such as some raptors and parrots, annual survival is often high (Newton et al. 2016; Maestri et al.2017) and individuals are long lived, but for small-bodied species like warblers and kinglets, rates of annual survival can be low (DeSanteet al. 2015; Johnston et al. 2016). While differences in body mass account for some of this variation ― larger species tend to live longer than smaller ones (Lindstedt & Calder 1976, 1981; Promislow 1993; Speakman 2005) ― many species live longer or shorter lives than predicted given their body mass (Healy et al. 2014). Other aspects of a species’ life history, particularly the demographic cost of reproduction, may explain this residual variation in survival rates (Williams 1966; Stearns 1992; Roff 2002). This view stems from the hypothesis that limited resources (i.e., time and / or energy) result in an allocation trade-off between two competing vital functions; specifically, current reproduction reduces future reproduction and survival. The pivotal survival-reproduction trade-off has been well documented in birds (Ricklefs 1977, 2000; Saether 1988; Linden & Møller 1989; Martin 1995; Ghalambor & Martin 2001), and with the observations of early investigators that the number of eggs laid declines from the poles towards the equator (Moreau 1944; Lack 1947; Skutch. 1949), it has given rise to the expectation that tropical species should offset a reduced clutch size by having higher adult survival (Murray 1985).
There are many studies that suggest high adult survival in tropical birds, the majority of which focus on comparisons between north-temperate systems and the tropics (Martin 2004). Early reports of high survival came from studies equating survival estimates with return rates (Snow 1962; Fogden 1972; Fry 1980; Bell 1982; Dowsett 1985). While these studies deepened our understanding of life-history strategies in tropical birds, estimates based on return rates are problematic because they confound estimation of complicated functions of survival rate and capture probability (Nichols & Pollock 1983; Krementz et al.1989; Sandercock 2006). More recently, studies employing improved methods for estimating survival via Jolly-Seber (JS) and Cormack-Jolly-Seber (CJS) models, which separate apparent survival (i.e., Φ: the product of true survival and site fidelity) from encounter probability (Sandercock 2006), have reinforced the idea of higher adult survival at lower latitudes (Faaborg & Arendt 1995; Johnston et al. 1997; Francis et al. 1999; Peach et al. 2001; McGregor et al. 2007). The generality of these findings, however, has been questioned based on comparisons showing negligible differences in survival between Central and North American birds (Karr et al.1990), and lower than expected survival rates for birds from South America (Blake & Loiselle 2008). Other studies have even found higher survival rates for south temperate birds compared to tropical species in Africa (Lloyd et al. 2014). Only one quantitative review has formally addressed latitudinal patterns in adult survival rates of birds across a broad range of latitudes. Muñoz et al. (2018) showed that adult survival was higher for species in the tropics compared to those in five sites across the north temperate zone, supporting the hypothesis of a latitudinal gradient in survival, at least for forest-dwelling passerines in the western hemisphere. Yet, despite longstanding interest in the idea of a latitudinal gradient in survival, we still lack an empirical synthesis at the global scale, which stands as a limiting factor in our ability to generalize these relationships to the diverse life history of birds found worldwide (Martin 2004).
Most explanations for a latitudinal survival gradient are based on the assumption of consistent latitudinal variation in survival and other life history traits with which it covaries, such as clutch size (Karret al. 1990; Faaborg & Arendt 1995; Johnston et al. 1997; Peach et al. 2001; McGregor et al. 2007). Indeed, most comparative studies of variation in life history traits treat northern and southern latitudes equivalently (Jetz et al. 2008; Muñozet al. 2018; Terrill 2018). However, this assumption may not always be met, since latitude itself does not directly influence organisms per se; rather, environmental factors that covary with latitude exert selective pressures on life history traits. For example, although there exists a global latitudinal gradient in clutch size (Cardillo 2002; Jetz et al. 2008), this trend is dampened in the southern hemisphere ― south temperate species lay smaller clutches than those in the north temperate hemisphere (Yom-Tov et al. 1994; Martin 1996; Evans et al. 2005). Consistent with this pattern, south temperate birds in Africa also tend to be longer lived than their north temperate European counterparts (Lloyd et al. 2014). This hemispheric asymmetry may in part be due to differing climatic conditions between northern latitudes and equivalent southern ones. Namely, south temperate latitudes are less seasonal and have higher minimum winter temperatures, both of which have been hypothesized to decrease adult mortality and lead to smaller clutch size (Ricklefs 1980). Similarly, clades and their intrinsic traits that may influence survival rates are also distributed nonrandomly across environmental gradients (Jetz et al. 2008; Sibly et al. 2012). Migratory habit, for instance, arises at least in part from species occupying higher latitudes and experiencing seasonal environments with lower minimum winter temperatures, and there can be substantial deleterious effects on survival over the migratory phase of the annual cycle (Sillett & Holmes 2002; Rockwell et al. 2017). Thus, the geographic variation in survival rates reflects a composite of extrinsic factors, intrinsic traits, and historical events related to a species’ lineage.
Because previous analyses of the latitudinal gradient in survival have focused on the north-temperate / tropical model (Martin 2004; Muñozet al. 2018) and have relied on a relatively narrow group of taxa, our current perspective of the biological underpinnings of the geographic variation in survival rates remains somewhat limited. Here, we present data on survival rates for 681 species of landbirds gathered from around the world (Fig. 1). The purpose of our analysis was to test the relative importance of latitude and extrinsic climate factors (temperature, precipitation, and seasonality) in explaining geographic patterns of avian survival rates, and to ask whether including intrinsic traits (body mass, clutch size, migratory habit) improved model predictions. Specifically, we ask: (1) Is there a latitudinal gradient in adult survival and, if so, are there differences between hemispheres? (2) Do climate measurements (extrinsic factors) explain differences in survival rates as well as latitude? (3) Do intrinsic traits explain additional variation in species-level survival rates? We tested for these relationships in both passerines and nonpasserines and between Old World and New World birds from mainland and island populations. By integrating data on macroecological processes with comparative biology, our modeling approach provides a powerful tool for understanding the diversity of life histories that have evolved across the globe.