Understanding population change in migratory songbirds: long-term and experimental studies of Neotropical migrants in breeding and wintering areas


*Email: Richard.T.Holmes@Dartmouth.edu.


Effective conservation and management of migratory bird species requires an understanding of when and how their populations are limited and regulated. Since 1969, my colleagues and I have been studying migratory songbird populations in their breeding quarters at the Hubbard Brook Experimental Forest in north-central New Hampshire, USA, and since 1986, in their winter quarters in the Greater Antilles (Jamaica). Long-term data on the abundance and demography of these populations, coupled with experimental tests of mechanisms, indicate that processes operating in the breeding area (e.g. density-dependent fecundity, food limitation) are sufficient to limit and regulate the local abundance of these species. At the same time, limiting factors operating in the non-breeding season (e.g. climate-induced food limitation in winter quarters and especially mortality during migration) also have important impacts on migrant populations. Furthermore, recent studies have shown that limiting processes during the winter period can carry over into the breeding season and affect reproductive output. These findings clearly demonstrate that to understand changes in abundance of long-distance migrant species requires knowledge of events operating throughout the annual cycle, which presents a challenge to researchers, managers and others concerned with the welfare of these species.

Despite the longtime interest and efforts of ecologists, the relative importance of the processes and factors determining the abundances of animals remain poorly understood (Lack 1954, Sinclair 1989, Murdoch 1994, Newton 1998). This is particularly true for migratory species that are distributed widely, often at global scales (Cox 1985, Sherry & Holmes 1995, Newton 2004a). Recent declines in the abundances of migratory songbirds in both North America (Robbins et al. 1989, Sauer et al. 2001) and Europe (Peach et al. 1998, Gregory et al. 2002, Newton 2004b, Sanderson et al. 2006) have raised anew the critical question of what limits or regulates these migratory songbird populations and when during the annual cycle these processes operate. One view is that migratory passerines are limited principally by events affecting survivorship in the non-breeding season (e.g. Baillie & Peach 1992, Morton 1992, Rappole & McDonald 1994). Alternatively, events on the breeding grounds, especially those affecting fecundity, have been implicated as being most critical for sustaining populations (e.g. Sherry & Holmes 1992, Böhning-Gaese et al. 1993, Holmes et al. 1996, Rodenhouse et al. 2003, Schmidt 2003, Sillett & Holmes 2005). Although evidence can be marshalled on both sides of this argument, theoretical considerations (Cox 1985, Svensson 1985, Sherry & Holmes 1995, 1996, Webster et al. 2002) and recent empirical findings (Marra et al. 1998, Sillett et al. 2000, Sillett & Holmes 2002, Norris et al. 2004a, 2004b, 2004c) support the hypothesis that habitat-mediated processes that are potentially limiting and/or regulatory probably occur in both winter and summer, as well as along migratory routes.

Focusing on this question of how, when and where migratory passerines are limited or regulated, my colleagues and I have been studying the ecology of Neotropical migrant songbirds in both their breeding grounds in temperate North America and their tropical wintering areas. Here, I provide a brief summary and overview of this research, which although it applies specifically to migratory passerines in North America, has broad applications to migrants elsewhere. In this paper, I consider (1) long-term trends in bird abundances at the Hubbard Brook Experimental Forest in New Hampshire, USA, (2) limiting and regulatory factors and processes operating on populations in both breeding and non-breeding areas, (3) survival across the annual cycle, (4) recent findings concerning interseasonal effects across different periods in the annual cycle and population connectivity, and (5) the implications of these findings for conservation and management of migratory bird populations.


Our breeding ground studies of Neotropical migrants have taken place at the Hubbard Brook Experimental Forest in the township of Woodstock, Grafton County, NH, USA (43°56′N, 71°45′W). The 3160-ha experimental forest is a component of, and contiguous with, the much more extensive White Mountain National Forest. The Hubbard Brook valley was extensively logged in the early 1900s, and the area was impacted by a hurricane in 1938. Vegetation in the study area at the time of this study consisted of uneven-aged, relatively mature, undisturbed and unfragmented second-growth northern hardwoods (Holmes et al. 1986, Holmes 1990, Holmes & Sherry 2001). The dominant tree species are Sugar Maple (Acer saccharum), American Beech (Fagus grandifolia) and Yellow Birch (Betula alleghaniensis), with an understorey of ferns, shrubs (particularly Hobblebush, Viburnum alnifolium) and small trees (e.g. Acer pensylvanicum, Acer spicatum, saplings of canopy species).

Bird species composition and the relative and absolute abundances of species occurring on the study area have varied markedly since 1969 when our studies began (Fig. 1; for details on individual species and their abundances, see Holmes & Sturges 1975, Holmes et al. 1986, Holmes & Sherry 1988, 2001, Holmes 1990). Approximately 15 bird species have bred annually on the long-term 10-ha study area, while 10–12 others have been present in some but not all years (Holmes 1990, Holmes & Sherry 2001). The majority of species and of individual birds in this forest are Neotropical migrants, i.e. those that migrate to tropical regions during the non-breeding period (Fig. 1). Two species that were abundant in the 1970s, the Least Flycatcher Empidonax minimus (maximum abundance = 57 individuals/10 ha in 1973) and Wood Thrush Hylocichla mustelina (maximum = 9 individuals/10 ha in 1977) subsequently disappeared from the study area, while a third, the American Redstart Setophaga ruticilla, declined dramatically (from a maximum of 44 individuals/10 ha in 1977 to 1–7/10 ha in the early 2000s). Largely because of changes in the abundance of these three species, the total number of individuals (all species combined) on the study area has declined from a high of 190–210/10 ha in the early 1970s to 75–90/10 ha in the late 1990s and early 2000s (see Fig. 1). A few species have increased in abundance since the study began (Ovenbird Seiurus aurocapilla, Black-throated Green Warbler Dendroica virens and Yellow-rumped Warbler D. coronata), while populations of most of the other breeding species, although fluctuating in abundance, have remained relatively stable over the 37-year period (Holmes & Sherry 2001, my unpubl. data).

Figure 1.

Thirty-seven-year trend in numbers of breeding birds occupying a 10-ha study area in northern hardwood forest at Hubbard Brook, New Hampshire, USA, 1969–2005. Numbers are totals for all species (males plus females) for permanent residents, short-distance migrants, long-distance (Neotropical) migrants and all species combined.


To identify limiting and regulatory processes in breeding areas, the demography of two parulid warbler species, the Black-throated Blue Warbler Dendroica caerulescens and the American Redstart were examined in depth. Long-term abundance data (1969–2005) for these two species are available from a 10-ha study area (Holmes et al. 1986, Holmes & Sherry 2001, my unpubl. data) and for 1986–98 on three replicate sites in other parts of the White Mountains (Holmes & Sherry 2001). Starting in the early 1980s, we expanded the Hubbard Brook study area to 64 ha and then to 100 ha for demographic studies of the two focal species. In these larger areas, we caught and individually marked with coloured rings all adults, and resighted returning individuals in subsequent years. We also followed and determined the fates of all nesting attempts, weighed nestlings prior to fledging, and estimated food (insect) availability and nest predator populations (see Holmes et al. 1992, 1996, Nagy & Holmes 2004, 2005a, 2005b, Sillett et al. 2004, Sillett & Holmes 2005). This information provided demographic information and data on important environmental factors. Because our data are most complete for the Black-throated Blue Warbler, the following discussion will focus on limitation and regulation of its population in the breeding period. This focal species occurs relatively commonly through forests in the northeastern United States and eastern Canada, and can be considered representative of the many small, forest-dwelling migratory passerines in temperate eastern North America.

Tests for density dependence

We found evidence for strong density dependence in the Black-throated Blue Warbler population at Hubbard Brook during the breeding season. Over the 37 years for which we have data, this Warbler population fluctuated from year to year, but remained relatively stable (Sillett & Holmes 2005, my unpubl. data). A time series analysis of abundances on the long-term census plot (1969–2005) showed strong density dependence (P = 0.0002, P.J. Doran & R.T. Holmes unpubl. data, see also Rodenhouse et al. 2003). Furthermore, fecundity (the number of young fledged/territory/breeding season) was significantly negatively correlated with adult Warbler density (Fig. 2A, Sillett & Holmes 2005). Recruitment of 1-year-old males into the population in the subsequent breeding season was also negatively correlated with adult warbler density (Fig. 2B, Sillett & Holmes 2005). No relationship was found between Warbler density and nest predation rate (Sillett & Holmes 2005). Using demographic data from our field population to parameterize a population model, we demonstrated that the observed density-dependent fecundity was sufficient to regulate this Warbler population (Sillett & Holmes 2005). The local abundance of this population therefore seems to be regulated by density-dependent processes, most related to factors affecting fecundity. Finally, not only was recruitment negatively related to adult Warbler density in the previous season, but there was also a strong and statistically significant positive relationship between recruitment and mean annual fecundity of Black-throated Blue Warblers in the previous breeding season (r = 0.78, P = 0.005, Sillett et al. 2000). This latter finding illustrates the importance of fecundity in maintaining local populations, even for a species that spends more than 8 months of the year away from the breeding grounds.

Figure 2.

Density-dependent relationships of Black-throated Blue Warblers at Hubbard Brook, New Hampshire USA: (A) annual fecundity (number of young fledged per territory per year) is negatively correlated with adult Warbler density in the same season (modified from Sillett & Holmes 2005), and (B) annual recruitment of yearling Warblers is inversely related to Warbler density in the preceding season (modified from Sillett & Holmes 2005). Numbers on the graph represent the years of study.

Determinants of fecundity

Demographic data gathered between 1986 and 1999, show that three factors – food availability (as indicated by changes in the Southern Oscillation Index, SOI; see below), nest predation and adult Warbler density – explained 91% of the variance in the mean number of young fledged/territory/year and 80% of the variance in the mean annual mass of young at fledging (Sillett & Holmes 2005). These findings led us to investigate how food, climate, nest predation and adult density influence Black-throated Blue Warbler fecundity.

First, food availability, as indicated by the abundance of Lepidoptera larvae – a major food item for our study species during the breeding season – varied greatly from year to year. Since studies at Hubbard Brook began in 1969, one major caterpillar irruption occurred in the early 1970s (Holmes et al. 1986, 1991). Since 1973, larval numbers have fluctuated annually, but at relatively low or endemic levels (Holmes et al. 1986, Reynolds et al. 2007). Between 1986 and 1998, these annual mean caterpillar abundances (as measured by larval biomass/unit vegetation) were significantly correlated with a climate variable – the mean monthly values of SOI, a standardized measure of the El Niño Southern Oscillation (ENSO, Sillett et al. 2000). Caterpillars were less abundant in El Niño years and higher in La Niña years. Mean annual Black-throated Blue Warbler fecundity and mean body mass of nestlings just prior to fledging were also correlated with SOI (r = 0.59, P < 0.04 and r = 0.79, P < 0.002, respectively), but adult Warbler density and annual nest predation rates were not (Sillett et al. 2000). Furthermore, fluctuations in the abundances of the six most common long-distance migrant species at Hubbard Brook and in three other sites in central New Hampshire were positively and significantly correlated with annual fluctuations in lepidopteran larvae (Jones et al. 2003). Thus, changes in the abundances of these species populations were synchronized with that of their lepidopteran food supply, which were related to the ENSO global climate patterns. These findings together suggest that weather, mediated in part by ENSO, influences caterpillar abundance (and biomass), which in turn affects Warbler fecundity, nestling growth and condition. Food is therefore implicated as an important factor affecting bird populations breeding in these temperate forests.

To test whether food was actually a limiting factor, we performed both food reduction and food augmentation experiments. Using a combination of observations across years and experimental reductions of caterpillar populations by aerial spraying with a larvacide, Bacillus thuriengensis (Bt), Rodenhouse and Holmes (1992) showed that in seasons with more abundant larvae when compared with years when food was scarce, both the number of young Black-throated Blue Warblers fledged per clutch and the number of clutches per season increased, while the frequency of nestling starvation decreased. No change occurred in clutch size across years or food treatments. Similarly, in experiments involving food supplementation rather than food reduction, females given more food fledged significantly more young per season compared with those that were not. In these experiments, the change in annual fecundity was not due to an increase in clutch size or the number of young fledged per nesting attempt, but to an increase in the frequency of multiple brooding by the females (Nagy & Holmes 2005b). Thus, food clearly limits reproductive output in this species, and this limitation probably occurs to at least some extent in most breeding seasons (Nagy & Holmes 2005b).

A second major factor shown to affect annual fecundity significantly was nest predation. Nest predation is a major source of mortality for Warbler eggs and young at Hubbard Brook (Holmes et al. 1992, my unpubl. data), as it is for many passerine species (Ricklefs 1969, Newton 1993). Annual predation rates on Black-throated Blue Warbler nests range from as low as 17% to a high of 42% (my unpubl. data). There is a large suite of nest predator species at Hubbard Brook, both bird and mammal (Reitsma et al. 1990, Sloan et al. 1998, my unpubl. data). Two sciurids – Red Squirrels (Tamiasciurus hudsonicus) and Eastern Chipmunks (Tamias striatus) – are the most important, but mice (e.g. Peromyscus spp.), raptors (Accipiter striatus), mustelids (Martes pennanti) and corvids (e.g. Cyanocitta cristatus) are also involved. Nest predation by snakes has not been documented at Hubbard Brook, nor has brood parasitism by Brown-headed Cowbirds (Molothrus ater). The latter species, however, does occur in human-disturbed habitats in valley areas a few kilometres away.

Both experimental studies (Reitsma 1992, Sillett et al. 2004) and analyses of field data (Sillett & Holmes 2005) indicate that the nest predation rate at Hubbard Brook is not related to nest density nor to adult Warbler density, i.e. it is not density dependent. The abundances of the sciurids, the major nest predators, are most strongly affected by seed crops of the dominant tree species, especially American Beech, which occur irregularly at 2–4-year intervals (my unpubl. data). Thus, year-to-year differences in predation rate on Black-throated Blue Warbler nests vary closely with the masting cycle of the tree species in this forest, and even though nest predation is an important source of mortality (i.e. an important limiting factor), it is not a major regulatory force.

The third major influence on annual fecundity was adult Warbler density, i.e. intraspecific density. The inverse relationship between adult density and the number of young fledged per female per season (Fig. 2A, see Rodenhouse et al. 2003, Sillett & Holmes 2005) is a classic example of density dependence. It, in combination with the effect of climate/food and nest predation, acts to limit and regulate the local abundance of this population during the breeding season (Sillett & Holmes 2005). Density of adult Black-throated Blue Warblers in this forest is further related to density of vegetation in the shrub layer, especially of Hobblebush, a major nesting and foraging substrate (Steele 1992, 1993, Doran & Holmes 2005). Finally, in our studies at Hubbard Brook, no evidence has been found for calcium as a limiting nutrient (Taliaferro et al. 2001), as has been reported for some passerines in Europe (e.g. Graveland & Drent 1997).

Mechanism(s) of density dependence

To explore the mechanisms underlying the density-dependent fecundity in this population, we tested two hypotheses: (1) crowding and (2) site dependence. A crowding mechanism would involve some interaction among close neighbours in a population that might interfere with breeding activity, leading to lower reproductive output. To investigate this mechanism experimentally, we reduced local density by removing conspecifics from around focal Black-throated Blue Warbler territories and then compared the behaviour and reproductive success of these focal birds with those on control territories (Sillett et al. 2004). The results indicated that young fledged per territory, territory size and the proportion of time males spent foraging were all significantly greater for birds on territories with fewer neighbours. The effect of neighbour removal was most pronounced in an El Niño year when conditions for breeding were least favourable. Thus, crowding can mediate interactions among territory holders that result in lower fecundity. Crowding therefore accounts for at least a part for the observed density dependence. In this case, crowding involved a change in male behaviour and use of space that may have its strongest impact when environmental conditions are relatively poor (Sillett et al. 2004).

Another mechanism that could generate the density-dependent negative feedback shown in Figure 2 involves what Rodenhouse et al. (1997) termed site-dependent regulation. This mechanism extends the ideas of Pulliam (1988), Pulliam and Danielson (1991) and Dhondt et al. (1992), and involves the pre-emptive use of sites that differ in suitability for survival or reproduction (McPeek et al. 2001). The basic premises are (1) that sites (territories) differ in quality, (2) that individuals occupy the best site available and are seldom replaced (pre-emptive occupancy), and (3) that site suitability determines reproductive success or survival. Tests of this mechanism conducted at Hubbard Brook (Rodenhouse et al. 2003, 2006) reveal that sites occupied by individual territorial birds differ strongly in quality, even in seemingly homogeneous habitat, that sites occupied every year are on average better than those occupied only periodically, and that the best sites (most food, fewest predators, highest vegetation density) yield the greatest annual fecundity. Similarly, Doran and Holmes (2005) reported that Black-throated Blue Warblers showed temporal and spatial variability in their choice of territory sites, but selectively chose territory sites with higher shrub density, greater food levels and lower predator levels, all of which influenced individual reproductive output.

Thus, heterogeneity among sites (territories) coupled with density-dependent territorial behaviour of male Black-throated Blue Warblers contributes to the annual variability in fecundity at the population level. Our findings indicate that both mechanisms – crowding and site dependence – are involved simultaneously in population regulation. Crowding operates in high-quality, high-density habitats at the scale of individuals and their neighbours, while site dependence occurs on a larger (i.e. landscape) spatial scale (Rodenhouse et al. 2003, 2006, Sillett et al. 2004).


One of the major problems in analysing long-term trends in species’ abundances and their causes is that the environment is continually changing over time, either due to natural processes such as succession and disturbance, or due to human-caused changes. In the relatively undisturbed and unfragmented forests at Hubbard Brook, three bird species that had relatively high densities in the early years of this study and that subsequently declined sharply, i.e. Least Flycatcher, Wood Thrush and American Redstart (see above), account for most of the decline in bird abundance on the long-term census plot (Holmes & Sherry 2001). These three species occur most commonly, reach their highest densities and probably have their highest reproductive output in early- to mid-successional forests (Holmes et al. 1986, Holmes & Sherry 2001). This suggests that the forest at Hubbard Brook had changed over time, leading these species to shift their breeding activities elsewhere. This is consistent with the idea that each bird species is affected by, and responds to, its environment in complex and often unique ways (Holmes et al. 1986, Holmes 1990). These three species, in particular, settle in forests at a certain stage of succession, in this case in early- to mid-successional stages approximately 20–60 years following major disturbance, such as forest harvesting and clear-felling. We have proposed that as the forest becomes older and changes occur in vegetation composition or structure, the site no longer remains suitable for these particular bird species and they decline in abundance (Holmes et al. 1986, Holmes & Sherry 1988, 2001). Other species, such as those that have increased during the 37 years of study, respond in different ways, and show preferences for the older, more mature forest stands. Such turnover of bird species along successional gradients is well known (e.g. Johnston & Odum 1956, May 1982), but has not been fully appreciated in recent assessments of population decline in Neotropical migratory songbirds (Litvitas 1993, Askins 2000).

This relationship of bird species to successional changes in vegetation structure was studied at Hubbard Brook and surrounding parts of New England by Hunt (1996, 1998). Comparing American Redstarts in early-successional habitats with those in more mature deciduous forest, Hunt (1996) found that individuals arrived earlier, established smaller territories and thus occurred at a higher density in early-successional forests than in more mature ones. There were also more older males in the earlier-successional sites, and these had higher mating success. Through a modelling effort incorporating historical patterns of vegetation change in the region over a period of decades, Hunt (1998) further showed that regional population trajectories of Redstarts were correlated with the amount of early-successional habitat present and that the recent decline in Redstart abundance reflected a loss of habitat of the appropriate age (and probably structure). Furthermore, Hunt suggested that mature forests may in effect function as second-best or even sink habitats for the American Redstart, its regional abundance being determined by reproductive output from earlier-successional sites.

Besides natural succession, changes in habitat structure and quality can also be affected by major disturbances, such as hurricanes and ice storms, and by anthropogenic changes, such as forest clearing and especially fragmentation. Habitat fragmentation, which has been studied extensively in parts of North America (see Faaborg 2002), leads to higher nest predation rates (see reviews by Chalfoun et al. 2002, Schmidt 2003) and to increased frequency of brood parasitism by Brown-headed Cowbirds, which greatly affect demographic rates (Robinson et al. 1995).

To summarize, our studies of migratory passerines on their temperate breeding grounds have shown that local abundance of our focal species is regulated by density-dependent fecundity, which is due in part to crowding and site-dependent mechanisms. Fecundity in turn is affected by multiple factors, including weather-induced food limitation (including the global climate change as illustrated by correlations with ENSO), nest predation and local density-related effects. Finally, changes on a longer temporal or wider regional scale are influenced by changes in habitat characteristics and availability, including those due to natural succession, disturbance and, in other regions outside of the Hubbard Brook forest, anthropogenic effects.


For assessment of factors affecting migrants in the winter, we have studied our two focal species, Black-throated Blue Warblers and American Redstarts, in their Neotropical winter grounds in Jamaica, West Indies, since 1986. Both species occupied a range of habitats from second-growth scrub to mangroves to wet forest (Lack & Lack 1973, Holmes et al. 1989), as well as agricultural habitats such as coffee and citrus (Johnson et al. 2006). Within habitats, individuals were dispersed through aggressive interactions. Black-throated Blue Warblers were most abundant in moist forests, where males and females usually held adjacent and spatially intermixed territories (Holmes et al. 1989). Redstarts were distributed across a greater array of habitats, and were most abundant in moist tropical forest, mangroves and shade coffee (Johnson et al. 2006). Male Redstarts were behaviourally dominant over females, and often excluded females from more preferred habitats (Marra et al. 1993), resulting in sexual habitat segregation (Marra 2000, Marra & Holmes 2001). Both species were strongly site faithful, both within winter seasons and among years (Holmes et al. 1989, Holmes & Sherry 1992).

Several lines of evidence from our studies suggest that food for migrant insectivorous birds is limiting in the winter period in Jamaica (see review by Sherry et al. 2005). First, Warbler abundance was correlated with differences in insect abundance across habitats in Jamaica, suggesting that food abundance was driving this pattern of distribution (Johnson & Sherry 2001). Comparisons of foraging behaviour of Redstarts in breeding and wintering areas showed that individuals in winter spent more of the daylight hours in foraging and used a great variety of often energy-costly techniques to capture prey, suggesting that prey were more difficult to find and capture in the winter period than in the summer (Lovette & Holmes 1995). Survival and body condition of Redstarts varied among habitats in Jamaica, with individuals in poor-quality sites (mostly females) not able to maintain body mass, which in turn led to delays in their spring departure from wintering sites (Marra 2000, Marra & Holmes 2001). Similarly, individuals in female-biased habitats had elevated baseline corticosterone levels and reduced acute corticosterone secretion compared with those in male-biased habitats, indicating more stressful conditions in the habitats to which females were relegated (Marra & Holberton 1998). More recently, Johnson et al. (2006) reported that the decline in mean body mass of American Redstarts over the winter period, which varied among habitats, was a strong predictor of apparent survival rates, and Brown and Sherry (2006) showed that food supply strongly controls body condition of another Neotropical migrant, the Ovenbird, during the winter period.

These findings of territorial dispersion, intraspecific behavioural interactions among and between the sexes, differences in physiological condition, and the potential for food limitation suggest that habitats in the winter period differ in quality and that migrant birds may often compete intraspecifically for the better sites. Even though these migrant species may occupy a wide range of habitats, the quality of those habitats differs markedly and can affect body condition and survival (see citations above, Johnson et al. 2006). Finally, these results indicate that habitat quality in the winter period may have important influences on populations (see below), and the loss of high-quality sites through deforestation, habitat degradation or climate change has major consequences for these migrant populations. Comprehensive data on habitat availability and habitat-specific demography in the winter period are lacking for most Neotropical migrants, but such information is needed for a full assessment of the importance of the winter period as a limiting season.


To determine survivorship, we used data from resightings of colour-ringed Black-throated Blue Warblers on breeding grounds at Hubbard Brook in New Hampshire and in winter quarters in Jamaica during the period 1986–2000 (Sillett & Holmes 2002). Individuals ringed in May at Hubbard Brook that were still present there in early August provided a measure of over-summer survival. Similarly, those present on winter sites in October that were resighted in March were used to estimate over-winter survival. Annual survival was calculated from resightings of birds from May to May (at Hubbard Brook) and from October to October (in Jamaica). Finally, by combining data on survival from the stationary (over-summer, over-winter) and migratory (March to October, August to May) periods, it was possible to estimate survival during the migratory periods (see Sillett & Holmes 2002).

Cormack–Jolly Seber models indicated that the survival probability of Black-throated Blue Warblers for the 3-month over-summer (May to August) period was 0.99, while that for the 6-month over-winter period (October to March) was 0.93. These values converted to survival probabilities per month were 0.99 and 0.99, respectively (Fig. 3). By contrast, the monthly survivorship rates calculated for the migratory periods ranged from 0.77 to 0.81 (Fig. 3). These findings indicate that apparent mortality rates were at least 15 times greater during migration than during stationary periods and that approximately 85% of apparent annual mortality of Black-throated Blue Warblers occurs during the migratory periods (Sillett & Holmes 2002). Thus, mortality during migration is of great importance to overall abundances of these migrant populations.

Figure 3.

Average monthly survival probabilities during each of four seasonal intervals in the annual cycle of the Black-throated Blue Warbler, a Neotropical migrant (modified from Sillett & Holmes 2002). ‘Summer’ = breeding season survival measured May–August at Hubbard Brook, New Hampshire, USA, and ‘Winter’ = over-winter survival measured October–March in Jamaica. Survival probabilities for the 6-week spring and autumn (fall) migratory periods were calculated from survival estimates measured in May and August in New Hampshire and in October and March in Jamaica, some of which spanned the migratory periods (see Sillett & Holmes 2002: 299).


Several lines of evidence suggest that events influencing migrants in one season may carry over and have effects on the population in the next season. An example at the population level is the finding that high annual fecundity in the summer period was correlated with higher numbers of hatch-year birds in the ensuing winter season and of yearlings in the next breeding season (Sillett et al. 2000). On an individual level, Marra et al. (1998), using stable isotopes to identify habitats occupied in winter, showed that Redstarts from high-quality winter habitat arrived earlier in spring at Hubbard Brook, and they hypothesized that this early arrival would affect reproductive success. Subsequently, Marra and colleagues provided confirming evidence that the reproductive success of Redstarts in their temperate breeding areas was influenced by the quality of habitats thousands of kilometres away on tropical winter quarters (Norris et al. 2004a, 2004b). Similarly, in a related study, Norris et al. (2004c), again using stable isotope analyses, demonstrated that breeding effort in Redstarts influences life-history strategies (e.g. location of moult) and the carotenoid concentration in feathers (e.g. affecting their signalling capabilities in a social context). These findings indicate that events occurring in one season can have profound effects on individuals and populations in subsequent seasons. Such interseasonal effects carrying over from one season to the next have just begun to be investigated (Norris 2005), and will require more intensive study and new methods of tracking individuals over the course of the annual cycle.


The findings reported above indicate that for a comprehensive understanding of population dynamics of long-distance migratory species, it is important to examine populations at all phases of the annual cycle. Ideally, this study of year-round ecology should involve the same individuals in all seasons. One caveat to our studies summarized above is that the birds we studied in New Hampshire were not the same individuals we studied in Jamaica (Sillett & Holmes 2002), although demographic linkages indicate that these are local samples of a larger breeding population (see above). Indeed, for most migratory species (but see Gunnarsson et al. 2005) and especially the smaller passerine species, such information on the relationship, if any, between local breeding and wintering populations is not available (Webster et al. 2002). For most passerines, ringing data and other direct marking procedures are insufficient to make these connections directly (Holmes & Sherry 1992, Webster et al. 2002), although telemetry is yielding useful information for large-bodied species (e.g. Ueta et al. 2000, Martell et al. 2001). Recent advances in the use of molecular markers (Smith et al. 2005), stable isotopes (Hobson 2005, Kelly 2006) or a combination of these methods (Clegg et al. 2003, Kelly et al. 2005) are helping to identify the geographical origins and dispersion of migratory birds.

To determine the connectivity between breeding and wintering areas for Black-throated Blue Warblers, we employed stable isotopes of hydrogen (deuterium) and carbon (13C) in feathers and other tissues (Chamberlain et al. 1997, Rubenstein et al. 2002, Royle & Rubenstein 2004). The results indicate that Warblers wintering in the western Greater Antilles (mainly Cuba and Jamaica) come primarily from the northern portion of the species’ North American breeding range (New Brunswick to Ontario south to New York state), while those in the more easterly Antilles (Hispaniola, Puerto Rico) breed mostly in the southern part of the breeding range, along the Appalachian mountain chain (Fig. 4). Furthermore, analysis of these stable isotope values indicates that individuals within local wintering sites, within our forest plots in Jamaica, for example, are derived from a wide range of breeding longitudes. Considerable mixing of individuals from various parts of the breeding range in local wintering sites is occurring, and conversely, individuals from local wintering sites are dispersing widely in the breeding grounds (Rubenstein et al. 2002). Consequently, loss of winter habitat would have a broad and diffuse effect on breeding populations and vice versa. Further studies examining the dispersion of breeding and wintering populations in subsequent seasons are urgently needed for future conservation and management planning.

Figure 4.

Connectivity between breeding and wintering populations of Black-throated Blue Warblers, as indicated by analyses of stable isotope values in feathers (Rubenstein et al. 2002). The pattern suggests that individuals from the northern portion of the breeding range migrate to and winter in the western Greater Antilles (mainly Cuba and Jamaica), while those breeding in the more southern Appalachian Mountains winter mostly in the more easterly Antilles (Hispaniola, Puerto Rico). Results from Rubenstein et al. (2002).


The results of our long-term studies of Neotropical migrant passerine birds in breeding and wintering areas and their implications can be summarized as follows:

  • 1Survivorship is affected by climatic variation, and especially by mortality occurring during migration. Events during migration and especially the quality of migratory stopover sites are important to the maintenance of long-distance migrant populations.
  • 2Fecundity is critically important in maintaining populations, and is affected by food availability, nest predators, climate and local density-dependent processes. Thus, maintenance of the quality of breeding habitat, the characteristics of which often vary for each species, is important.
  • 3Habitat quality changes over time as succession proceeds, and through the processes of disturbance, both natural and human-caused. Moreover, habitat quality varies among species, with some preferring one particular habitat type or habitat state and not others. Management plans for creating or maintaining suitable habitat therefore must take into account which species are most in need of consideration and their particular preferences and requirements.
  • 4Finally, and most importantly, population abundance is determined by a complex interaction of biotic and abiotic factors operating in both breeding and non-breeding seasons. Thus, to understand population change in migrant species, it is necessary to consider population dynamics and the factors that affect them through the annual cycle – in breeding grounds, wintering quarters and along migratory routes. As individuals of migratory species move long distances between breeding and wintering areas and occupy a variety of habitats over the course of the annual cycle, this presents a major challenge to researchers, conservationists and land managers concerned with the welfare of these populations.


Many colleagues, students and field assistants have been involved in important and invaluable ways to this research over the past 37 years. I especially thank P.J. Doran, P.D. Hunt, P.P. Marra, L.R. Nagy, N.L. Rodenhouse, T.W. Sherry and T.S. Sillett for their many contributions to this project. N.L. Rodenhouse and T.S. Sillett read and commented on the manuscript, and T.S. Sillett and D. Rubenstein helped in preparation of the figures. The studies at Hubbard Brook were conducted under the auspices of the Northern Research Station, Forest Service, US Department of Agriculture, while the studies in Jamaica would not have been possible without the support of the Sutton and Williams families and permission given by the Natural Resources Conservation Authority. Finally, the US National Science Foundation has funded this research since its inception.