Site Fidelity of Shrubland and Forest Birds

Abstract.

In eastern North America, most early-successional woody habitats are ephemeral and succeed to forests within a few decades. Consequently, for shrubland birds patches of habitat are generally suitable for breeding for only a short time. This has led some authors to suggest that shrubland birds should show little fidelity to former breeding sites and frequently disperse to new, more suitable patches. Similarly, one would expect birds of mature forests to have greater site fidelity because their habitats are more stable. Understanding the dispersal strategies of shrubland birds has important implications for the management and conservation of these declining species. Therefore, I used meta-analysis to compare rates of site fidelity of migratory shrubland and forest passerines that breed in eastern North America. For adult birds, site fidelity of shrubland (estimate ± SE = 0.353 ± 0.022) and forest species (0.364 ± 0.036) was similar. Philopatry of yearling birds to their natal site was lower but similar for shrubland (0.048 ± 0.014) and forest birds (0.028 ± 0.019). I conclude that dispersal rates of shrubland birds are not unusually high. Rather, both forest and shrubland species appear to follow decision rules that are common to passerines, with adults frequently returning to former breeding sites and yearlings dispersing away from their natal sites. For management, these results are encouraging, as populations of shrubland birds should persist in highquality habitats maintained in an early-successional state.

Resumen.

En el este de Norte América, la mayoría de los ambientes leñosos en estadíos tempranos de sucesión son efimeros y se transforman en bosques en pocas décadas. En consecuencia, los parches de hábitat son generalmente adecuados para la reproducción de las especies de aves de arbustales sólo por un tiempo corto. Esto ha llevado a algunos autores a sugerir que las aves de arbustales deberían exhibir una baja fidelidad a sus sitios de reproducción previos, y deberían dispersarse frecuentemente a parches nuevos y más adecuados. De modo similar, se esperaría que las aves de bosques maduros presenten mayor fidelidad a los sitios debido a que sus hábitats son más estables. Entender las estrategias de dispersión de las aves de arbustales tiene implicaciones importantes para el manejo y la conservación de estas especies en declive. Por lo tanto, utilicé meta-análisis para comparar las tasas de fidelidad a los sitios de aves paserinas migratorias de arbustales y de bosques que se reproducen en el este de Norte América. Considerando aves adultas, la fidelidad a los sitios de las aves de arbustales (estimado ± EE = 0.353 ± 0.022) y de bosques (0.364 ± 0.036) fue similar. La filopatría de las aves de un año de edad a su sitio natal fue menor, pero similar en las aves de arbustales (0.048 ± 0.014) y de bosques (0.028 ± 0.019). Concluyo que las tasas de dispersión de las aves de arbustales no son inusualmente altas. En cambio, tanto las especies de aves de bosques como las de arbustales parecen seguir reglas para tomar decisiones que son comunes entre los paserinos, de manera que los adultos regresan frecuentemente a los sitios de reproducción y las aves de un año de edad se dispersan alejándose de sus sitios natales. Para propósitos de manejo, estos resultados son esperanzadores ya que las poblaciones de aves de arbustales deberían persistir en ambientes de alta calidad mantenidos en estadíos sucesionales tempranos.

Introduction

The habitats used by shrubland birds in eastern North America are largely ephemeral (Askins 2000, DeGraaf and Yamasaki 2003). In regenerating clearcuts, the most common shrubland habitat, shrubland birds are abundant for the first 10–15 years after logging (DeGraaf 1991, Keller et al. 2003). Because of succession, however, by the time a clearcut reaches 20 years of age, nearly all shrubland birds have disappeared (Schlossberg and King 2009). Thus habitats for shrubland birds exist in a “shifting mosaic” (Lent and Capen 1995). Each year, older areas of scrub become unsuitable as succession progresses, while new patches of shrubland are created by logging or other disturbances. Even if the extent of shrubland habitat in a region is constant, locations of suitable patches likely change from decade to decade.

Because shrublands are ephemeral, one might expect shrubland birds to adopt dispersal strategies different from those of birds in more stable habitats like deciduous forests. If shrubland patches remain suitable for only a short time, then returning to a previously used site could be risky, and changing breeding sites frequently could be adaptive (Switzer 1993). In larids, site fidelity, the tendency to return to sites of previous breeding, increases with the temporal stability of the nesting habitat (McNicholl 1975). Because shrubland birds occupy ephemeral habitats, one might expect these species to have lower site fidelity than birds of more stable habitats. In fact, some authors have suggested that shrubland birds are “fugitive” species (Hutchinson 1951) that move freely from patch to patch between years (Askins 1993, 2000, Lent and Capen 1995). Forests, unlike shrublands, are relatively stable, so similar logic would lead one to predict that forest birds return to former breeding sites at higher rates than shrubland birds.

With shrubland birds in serious decline in the eastern U.S., understanding their dispersal strategies has important conservation implications (Askins 1993, Hunter et al. 2001, Schlossberg and King 2007). If shrubland birds have low site fidelity and disperse frequently, then populations in any location might fluctuate widely from year to year, and typical management strategies of creating and restoring habitat may be ineffective because populations are not necessarily concentrated in high-quality habitat. Simberloff and Abele (1976) suggested that dispersal-prone species might be better managed with many small reserves than a single large one.

On the other hand, if shrubland birds return to their previous breeding sites frequently, then populations will be more closed, and management will be more straightforward (Robinson and Morse 2000). High site fidelity, however, could create a mismatch between birds' evolved behavioral strategies and their anthropogenic habitats. Before humans arrived in North America, shrubland birds likely used habitats such as beaver ponds, floodplains, coastal heaths, and pitch pine—scrub oak woodlands (DeGraaf and Miller 1996, Askins 2000). Because of edaphic conditions, these habitats may have been relatively stable over time (Latham 2003). Today, in contrast, regenerating clearcuts are the largest source of shrublands in the eastern U.S. (Schlossberg and King 2007). New clearcuts rapidly grow into forests and provide shrubland habitat for <20 years (DeGraaf and Yamasaki 2003, Schlossberg and King 2009). Thus, for shrubland birds, high site fidelity to very ephemeral habitats could be maladaptive and even create an ecological trap if the habitat's quality declines during succession.

To date, there has been no synthesis of dispersal strategies of shrubland birds. I conducted a meta-analysis of studies reporting site fidelity of these birds. My goal was to determine how frequently shrubland birds of eastern North America return to former breeding sites. To provide a yardstick against which to compare results for shrubland species, I also assessed return rates of birds of mature forests.

Methods

I used meta-analysis to compare site fidelity of migratory shrubland and forest passerines that breed in the eastern U.S. and southeastern Canada, defined as east of the Mississippi River in the U.S. and east of Manitoba in Canada. My list of shrubland species was based on Schlossberg and King (2007), Hunter (2001), and AOU (1998). For forest birds, I based my species list on DeGraaf and Yamasaki (2001) and AOU (1998), but I excluded species associated primarily with disturbance-dependent habitats such as savannas of longleaf pine (Pinus palustris), as I was interested in birds found in relatively stable habitats. In the eastern U.S. most shrubland birds are migratory (Schlossberg and King 2007). For nonmigratory passerines, the concept of site fidelity is elusive because dispersal distances are short (Paradis et al. 1998). Therefore I focused exclusively on migratory populations and excluded studies of populations whose migration was described as partial, facultative, or altitudinal, as not all birds in such populations may actually migrate.

For each shrubland and forest species, I searched the literature for studies in which researchers individually marked breeding birds, nestlings, or fledglings during one or more breeding seasons and counted birds that returned in subsequent years. To search for these studies, I used ISI Web of Knowledge, Biological Abstracts, and references cited in the literature. I also used unpublished data reported in the Birds of North America series.

The raw data I compiled for the meta-analysis was the proportion of birds marked in one breeding season that returned to the same study area the following season. To avoid pseudoreplication, if a study followed marked individuals for multiple years, I included returns from only the first year after an individual was banded. Where studies reported results for different years or study sites separately, I summed results. In birds, the sexes tend to differ in their site fidelity (Clarke et al. 1997), and the studies I reviewed included either males only or both sexes. As a result, estimates of site fidelity by habitat may have been biased by how many studies included females. To overcome this bias, I controlled for sex in the analysis.

Past research suggests that natal philopatry in migratory birds is low, averaging just 0.039 across several habitat types (Weatherhead and Forbes 1994). This low rate of return could reflect a constraint on the ability of yearling birds to return to their natal areas or a strategy to prevent inbreeding by dispersing away from relatives. Whatever the cause, decision rules determining site fidelity appear to be different for yearling and older birds, with different selective pressures on each (Greenwood and Harvey 1982). Consequently, I analyzed site fidelity of returning yearling and older (hereafter “adult”) birds separately.

I analyzed data by means of mixed models that included a fixed effect of breeding habitat (forest vs. shrubland) and a random effect of study. In studies of site fidelity, factors such as plot size, habitat quality, or resighting effort can affect counts of returning birds (e.g., Bollinger and Gavin 1989, Cilimburg et al. 2002, Marshall et al. 2004). Ideally, I would have controlled for these variables, but only a few studies in my sample reported the relevant information. Therefore, I included a random effect of study to control for differences among studies. Because only one study in my sample included birds of both forest and shrublands, the study effect could theoretically account for all of the variation due to habitat. Results of the analysis, however, were nearly identical with and without the study effect, so I report results with the random effect included. For adult birds, I included a fixed effect of sex and an interaction between sex and breeding habitat to determine if differences between sexes were consistent across habitats. Most studies of natal philopatry did not report results by sex, so I did not include sex in models for yearling birds. Because studies with larger sample sizes produce better parameter estimates, I weighted studies on the basis of the number of birds marked (Lipsey and Wilson 2001).

I estimated site-fidelity rates only for birds that breed in eastern North America, as defined above. Some of these species, however, breed throughout North America, and many of the studies found in my literature search took place in western North America, where habitat types and disturbance regimes can be very different from those of eastern North America (Askins 2000). Thus, I conducted two separate analyses, one including all studies and one including only studies conducted in eastern North America.

I considered two possible confounding factors (see Weatherhead and Forbes 1994). First, study duration could affect estimates of return rates because samples from one or a few years might reflect atypical conditions rather than longterm averages (e.g., Trine 1996). Second, in some studies of natal philopatry, researchers captured recently hatched nestlings, while in others they captured fledglings in mist nets. This difference is problematic because fledglings could be dispersers prospecting for future breeding sites, and prospectors may have higher site fidelity than birds hatched on a site (Morton et al. 1991). I used linear regression to test the effect of study duration on site fidelity and used t-tests to evaluate the effect of capture method on site fidelity of yearling birds. Where I found significant effects, I incorporated them into the mixed models.

Observed rates of return are actually the product of three variables: resighting probability, annual survival, and site fidelity. As a result, observed effects of habitat on return rates may reflect variation in annual survival or resighting probability rather than site fidelity (Martin et al. 1995). For migratory birds, annual survival and site fidelity are difficult to separate because permanent dispersal is indistinguishable from mortality (Pollock et al. 1990). Resighting probability, however, can be estimated with mark—recapture methods (Lebreton et al. 1992). To determine whether differences in resighting probability affected results of the meta-analysis, I searched the literature for estimates of annual (apparent) survival of adults of all species for which I estimated site fidelity. I compared survival estimates of forest and shrubland birds with two-group t-tests. Throughout the results, I present parameter estimates as mean ± SE.

Results

Results of the meta-analysis were nearly identical whether I included all studies or only studies taking place in the East, so I report results for the full dataset. For adult birds, I located 60 studies reporting site fidelity for 38 species and a total of 15 426 marked birds (Table 1). Study duration had no effect on estimated rates of site fidelity (r² = 0.01, P = 0.27). Males had higher site fidelity than females in 32 of 35 studies with data on both sexes. After I controlled for sex differences, rates of site fidelity for shrubland (estimate = 0.353 ± 0.022) and forest birds (0.364 ± 0.036; Table 1) were similar. In the mixed model, habitat had no significant effect on site fidelity (Table 2). Similarly, for adults, apparent survival estimates of shrubland (0.473 ± 0.033) and forest birds (0.482 ± 0.046; t₂₆ = 0.2, P = 0.87) did not differ.

For yearling birds, I located 30 studies reporting site fidelity for 28 species and a total of 26 063 marked birds (Table 3). Site fidelity was similar whether birds were originally captured as fledglings or as nestlings (t₂₆ = 0.8, P = 0.44), and study duration had no effect on site fidelity (r₂ = 0.01, P = 0.62). Estimated natal philopatry was 0.048 ± 0.014 for shrubland birds versus 0.028 ± 0.019 for forest birds (Table 3). This difference was not significant (Table 2).

Discussion

Return rates of yearling and adult shrubland birds were similar to those of forest birds and passerines in other habitats (Greenwood and Harvey 1982, Weatherhead and Forbes 1994). This result held true even after I accounted for resighting probability. Therefore, I reject the hypothesis that shrubland birds have high dispersal rates because of their ephemeral habitats. Rather, shrubland birds appear to follow the same decision rules as birds in other habitats. Estimated survival rates of adult shrubland birds averaged 0.47 across several studies. Thus, a mean return rate of 0.35 suggests that approximately three-fourths of surviving adults return to former breeding sites. Like birds of other habitats, adult shrubland birds may use the win-stay, lose-shift strategy, showing site fidelity in locations where they breed successfully and dispersing when they fail to breed (e.g., Nolan 1978, Greenwood and Harvey 1982, Haas 1998, Hoover 2003). Yearling birds from both shrublands and forests generally disperse away from their natal sites to new breeding locations.

Why would adult shrubland birds return to previous breeding sites if such sites grow unsuitable after a few years because of succession? I offer three possible explanations. First, dispersing to a new breeding site may impose costs on a bird (Beletsky and Orians 1991, Bensch and Hasselquist 1991). Given the limited availability of shrublands, searching for a new territory will require time and energy and potentially expose birds to predation (Yoder et al. 2004). Even when dispersers find new breeding sites, those birds may be relegated to lower-quality territories by dominant, sitefaithful birds (Ward and Weatherhead 2005). Conversely, returning to a previously used site may confer advantages of site dominance (Bruinzeel and van de Pol 2004). For shrubland birds, returning to a previous site that is decreasing in quality because of succession may be more profitable than seeking a new site.

Second, even for shrubland birds that specialize on specific stages of succession, early-successional habitats can provide suitable habitat for a relatively long time. Regenerating clearcuts, for instance, are suitable for shrubland birds for 10 to 20 years after logging (Schlossberg and King 2009). This period is significantly longer than the lifespans of most small passerines (Klimkiewicz et al. 1983). Thus, returning to former breeding sites may be adaptive if the habitat is likely to remain suitable through a bird's expected lifespan. Interestingly, birds that breed in the grasslands of the Great Plains appear to show low site fidelity (Jones et al. 2007). These birds may have evolved a strategy of frequent dispersal and low site fidelity in response to annual fluctuations in habitat quality caused by large grazers, fire, or drought (Andersson 1980).

The above two rationales for high site fidelity of shrubland birds suggest that this behavioral strategy is adaptive. A third possible explanation is that this behavior is not adaptive but, rather, is an evolutionary relict. As discussed above, the natural shrubland habitats with which shrubland birds evolved may have been more stable than the clearcuts and other anthropogenic habitats used today. High site fidelity may have been adaptive in stable habitats of the past, but using such a strategy in more ephemeral habitats could be problematic. For example, in New Hampshire adult Chestnut-sided Warblers (Dendroica pensylvanica) showed high site fidelity in maturing clearcuts where habitat suitability was declining, as measured by overly mature vegetation and increasing rates of nest pre dation (D. I. King, pers. comm.). These birds returned to maturing clearcuts for several years after recruitment of new breeders had stopped. High site fidelity of shrubland birds may, therefore, be a maladaptive relict of past times, but more research is needed to test this hypothesis.

Interestingly, in my sample two species, one forest and one shrubland, showed very low site fidelity. Across three studies the Cedar Waxwing (Bombycilla cedrorum) had a mean return rate of just 0.01, and in one study the Rubycrowned Kinglet (Regulus calendula) had a return rate of 0.04 (the negative estimate from the mixed model is a consequence of the kinglet's site fidelity being dramatically lower than that of any other forest species). Cedar Waxwings choose breeding habitats on the basis of the availability of fruit, their primary food (Witmer et al. 1997). Because areas with fruit are patchily distributed and may vary in location from year to year, waxwings wander widely in search of suitable habitat. Populations of Ruby-crowned Kinglets appear to fluctuate widely from year to year, for unknown reasons (Swanson et al. 2008). These findings suggest that some bird species may have adopted a nomadic lifestyle even in habitats that superficially appear stable. The vast majority of shrubland and forest birds, however, appear to show greater fidelity to former breeding sites.

The tendency of yearling shrubland birds to disperse to new breeding sites, distant from where they hatched, was also found for forest species and appears to reflect a pattern widespread in passerines (Weatherhead and Forbes 1994). Because yearlings have never held a territory, they may not benefit from site dominance as site-faithful adults do. Returning to natal sites, therefore, may not provide yearlings any advantage, and dispersal may be favored to ensure outbreeding (Weatherhead and Forbes 1994). Support for this hypothesis comes from the fact that many migratory passerines show little genetic structure on all but the largest scales (e.g., Ruegg and Smith 2002, Clegg et al. 2003, Davis et al. 2006).

Given that many shrubland birds are declining, their strategies for site fidelity have important conservation implications. As discussed above, in clearcuts high site fidelity could be maladaptive. On the other hand, the combination of high site fidelity of adults and low site fidelity of yearlings may be beneficial in managed habitats that are maintained in an early-successional state. Populations should be expected to persist at sites where birds breed successfully, with newly arriving yearlings replacing former breeders that die or disperse (Robinson and Morse 2000). Thus conventional management prescriptions, such as restoring and maintaining habitat and controlling nest predation and brood parasitism, should build lasting populations of shrubland birds.

At the same time, however, important questions about dispersal in shrubland birds remain unanswered. One critical question is the distance over which birds, especially yearlings, disperse (Robinson and Morse 2000). If yearlings disperse only short distances, then populations will be effectively closed on larger scales even if they are open on a patch basis. Thus dispersal distance has important implications for the persistence of populations on regional scales (see Kinlan et al. 2005). Another unresolved issue concerns colonization of newly created shrublands by dispersers. My findings here suggest that most birds searching for new habitats should be yearlings. In some species, yearlings prefer to settle on patches already occupied by breeding birds (Ward and Schlossberg 2004). This behavior, conspecific attraction, can lead to otherwise suitable sites going unoccupied, which could affect population persistence negatively (Ray et al. 1991). The extent of conspecific attraction in shrubland birds is largely unknown. These examples show that knowing return rates of shrubland birds is only an initial step toward attaining a working understanding of their population dynamics.

The idea that shrubland birds have high dispersal rates is just one of several myths about these species that have been dispelled in recent years. In the past, shrubland birds were considered habitat generalists. Recent research, however, suggests that many shrubland birds are specialized in their habitat usage (Bulluck and Buehler 2006, Fink et al. 2006, King et al. 2009). Shrubland birds have also been considered edge species (Imbeau et al. 2003), but field data show that these birds avoid edges (Schlossberg and King 2008). These findings illustrate the need to test our conceptions about birds with field data lest we make improper decisions about management and conservation.

Acknowledgments

Thanks to D. King and two anonymous reviewers for excellent comments that helped me to improve the manuscript. Thanks to the Natural Resources Conservation Service, Resource Inventory and Assessment Division for financial support.

Literature Cited