A fundamental issue in population ecology lies in understanding the processes that limit or regulate animal populations (Newton 1979, 1998). Our understanding of these processes has been illuminated by several long-term studies on raptors, particularly on those populations that recovered after strong human-induced declines associated with persecution or DDT poisoning (e.g., Newton and Wyllie 1992, Ferrer and Donazar 1996, Nicoll et al. 2003, Bretagnolle et al. 2008, Watts et al. 2008, Elliott et al. 2011).

Population growth and dynamics may be constrained or affected by many different factors, both biotic and abiotic, which may affect available resources in the environment and competition levels for these resources (Newton 1979). Density-independent factors, such as climate, can influence demographic rates, and thus population growth and distribution patterns (Caughley 1988, Garcia and Arroyo 2001, Redpath et al. 2002). Additionally, if competition for limiting resources such as food, breeding territories, nesting sites, or mates occurs, the degree of competition typically increases with population size or density (e.g., Carrete et al. 2006a, 2006b, Bretagnolle et al. 2008). Density therefore often affects demographic parameters such as reproduction, emigration probability and survival rates, and thereby rates of population growth or decline.

Long-term studies of large raptors are relatively rare, because of the logistic constraints of monitoring large areas over many years. Despite potential methodological limitations (Balbontín and Ferrer 2008, Carrete et al. 2008), long-term monitoring projects remain extremely valuable for understanding the processes limiting or regulating populations. Understanding whether density-dependence occurs, with which demographic parameters and by which mechanisms, is crucial not only for understanding population dynamics but also for conservation (e.g., Carrete et al. 2006a).

In raptors there is broad evidence that population density influences breeding performance (Houston and Schmutz 1995, Ferrer and Donazar 1996, Nicoll et al. 2003, Bretagnolle et al. 2008, Elliott et al. 2011), and also some evidence that density influences survival (e.g., Nicoll et al. 2003, Serrano et al. 2005), thereby potentially contributing to the regulation of stable populations around a certain carrying capacity. However, the density-dependence relationships are complex and variable (Penteriani et al. 2006) and the mechanisms underlying the density-dependence relationships are still debated (Balbontín and Ferrer 2008, Carrete et al. 2008). The first steps consist of looking for evidence of population regulation, and determining the relative importance of density-dependent and density-independent factors (such as climate) in explaining variation in population growth rate and key demographic parameters such as reproduction and survival.

The Bald Eagle (Haliaeetus leucocephalus) is a large diurnal fish-eating raptor that occurs throughout the U.S.A., Canada, and northern Mexico (Gerrard and Bortolotti 1988, Buehler 2000). It is found near large bodies of open water with an abundant food supply and mature trees for nesting. Its diet consists mainly of fish, but it is also an opportunistic feeder (Gerrard and Bortolotti 1988, Dzus and Gerrard 1993, Buehler 2000). In Canada, except for British Columbia and Nova Scotia, it is mainly a breeding summer visitor, leaving after breeding to winter in the U.S.A.

In the mid- to late 1960s, the Bald Eagle became rare in the contiguous United States, mainly due to persecution by humans and negative effects of pesticides (DDT) on reproduction (Buehler 2000). The species was listed for protection under the Bald Eagle Protection Act in 1940, and the entire Bald Eagle population in the United States was listed for protection in 1978 under the Endangered Species Act of 1973. From the 1970s and 1980s, many Bald Eagle populations recovered following reductions in human persecution and DDT use (Kirk and Hyslop 1998, Buehler 2000, Watts et al. 2008). The species was removed from the U.S. federal government's list of endangered species (transferred to the list of threatened species) in 1995, and was completely delisted in 2007. In Canada, the Bald Eagle is currently considered “Not-At-Risk,” and as “secure” in the province of Saskatchewan (G Rank: G5; S Rank S5B, S4M, S4N; Saskatchewan Conservation Data Center 2012). There have been many studies on aspects of its breeding biology and population studies in the areas where the species had been decimated but recovered (Steidl et al. 1997, Buehler 2000, Elliott et al. 2005, Watts et al. 2008, Elliott et al. 2011). However, fewer studies have evaluated longer term population changes in the species' core populations (Gerrard et al. 1992, Dzus and Gerrard 1993), or the density-dependent and density-independent factors affecting those populations (Watts et al. 2008, Elliott et al. 2011).

In this study, our aim was to evaluate long-term changes in population size and breeding performance of Bald Eagles, using 45 yr of data from a nesting population located in the center of its breeding distribution in Canada (Besnard Lake, Saskatchewan). We also assess the relative importance of density-dependent (local population size) and of density-independent (local climate) factors in explaining variation in breeding performance and population change.

Methods

Results

Trends in Population Size and Reproduction

Throughout the study period (1968–2012), the number of occupied breeding areas at Besnard Lake initially increased during the 1970s and then stabilized (Fig. 1a). Using piecewise regressions, we identified two separate linear trends with a cut-off point in 1988 (±2.5 yr; NLIN procedure; F_3,41  =  22.46; P < 0.001). The eagle population increased until 1988 (slope ± SE: 0.581 ± 0.103) and remained relatively stable afterwards (slope: −0.077 ± 0.084; Fig. 1a). Changes in the number of successful pairs also followed this pattern. Piecewise regression identified an earlier cut-off point in 1977 (±1.9 yr; NLIN; F_3,40  =  9.02; P < 0.001). The number of successful pairs increased until 1977 (slope: 0.855 ± 0.261), but stabilized or slightly declined between 1977 and 2011 (slope: −0.057 ± 0.041; Fig. 1a). Annual production showed a similar initial increase until 1981 (±1.9 years; NLIN procedure; F_3,40  =  11.07; P < 0.001; slope: 1.046 ± 0.258), followed by a slow decrease between 1981 until 2011 (slope: −0.285 ± 0.082; Fig. 1b). Throughout the study period (1968–2011), nesting success declined linearly (Genmod: F_1,33  =  12.10; P < 0.01; slope: −0.023 ± 0.007; Fig. 1c).

Figure 1

Changes over time (1968–2012) in the (a) population size (number of occupied breeding areas [open circles] and number of successful breeding pairs [filled circles]), (b) annual production (black triangles) and (c) nesting success of Bald Eagles at Besnard Lake 1968–2012.

Density-dependence

Population growth rate (Log N_t+1/N_t) was negatively related to population size (N_t; F_1,42  =  31.47; P < 0.001; slope: −0.026 ± 0.005; R²  =  0.428; Fig. 2a). Similarly, mean productivity was significantly negatively related to population size (F_1,42  =  20.09; P < 0.001; slope: −0.028 ± 0.006; R²  =  0.323; Fig. 2b): the more eagles occupying territories, the fewer young they produced per pair. Both the nesting success (Genmod: F_1,42  =  11.67; P  =  0.006; slope: −0.056 ± 0.017) and mean brood size at fledging were negatively correlated with population size (GLM: F_1,42  =  9.44; P  =  0.004; slope: −0.015 ± 0.005; R²  =  0.183; Fig. 2b). If we excluded one outlier (for the year with a population size of 8; Fig. 2b), similar relationships were found between population and productivity (F_1,41  =  12.61; P < 0.0001; slope: −0.0262 ± 0.007), nesting success (F_1,41  =  9.15; P < 0.001; slope: −0.053 ± 0.018) and mean brood size at fledging (F_1,41  =  8.29; P < 0.001; slope: −0.0168 ± 0.006).

Figure 2

Density-dependence of the (a) population growth rate from year t to t+1 and (b) reproduction in year t (mean productivity [open circles] or mean brood size at fledging [filled circles]) of Bald Eagles at Besnard Lake. On the X-axis, population size refers to the number of occupied breeding areas in year t.

Considering the number of failed nesting pairs in a given year (the number of occupied breeding areas without success), there was an even stronger negative relationship between mean brood size at fledging and numbers of failed nesting pairs (GLM: F_1,42  =  10.44; P  =  0.002; slope: −0.021 ± 0.006; R²  =  0.200). Thus, when there were more unsuccessful nesting eagles within the population, the successful pairs produced fewer young.

Fluctuations in Breeding Performance

After controlling for the effect of density on reproduction parameters (using the residuals from GLMs of mean productivity, nesting success or mean brood size at fledging on population size, N), we found evidence for autocorrelation with a 5-yr time-lag for the relative productivity (Durbin-Watson test; DW  =  1.142; P  =  0.006 for an order of 5). For the relative nesting success, the autocorrelation was significant for time-lags of 5 yr (DW  =  0.989; P < 0.001 for an order of 5) and 10 yr (DW  =  1.091, P  =  0.037 for an order of 10; Fig. 3). No significant autocorrelation was detected for the relative mean brood size at fledging.

Figure 3

(a) Changes over time (1968–2011) in the relative nesting success of Bald Eagles (nesting success corrected for population size; see text) and (b) auto-correlogram for this variable, showing a 5-yr period (***: indicates a significant correlation at the P < 0.001 level; *: indicates a significant correlation at the P < 0.05 level).

Discussion

Analyses of Bald Eagle population trends on Besnard Lake, Saskatchewan, showed a period of increase during the 1970s and 1980s, followed by a relatively stable phase (Fig. 1a). In trying to understand these changes in population dynamics, we considered factors influencing eagles on their wintering grounds in the midwestern U.S.A. and on their breeding grounds of northern Saskatchewan. In 1940, the United States passed legislation to protect Bald Eagles, after which there was a gradual increase in the number of Bald Eagles wintering in the midwestern U.S.A. until at least the early 1970s (Buehler 2000, Steenhof et al. 2002). In the 1970s and 1980s, Bald Eagle populations in the western Great Lakes were recovering from DDT (Buehler 2000). Unlike most other North American populations (Bowerman et al. 1998, Watson et al. 2002, Watts et al. 2007), the Bald Eagle population of Besnard Lake was unlikely to have been affected by DDT, as DDT was not used in Saskatchewan's boreal forests and Saskatchewan Bald Eagles tended to winter in areas which were less contaminated by DDT (Gerrard and Bortolotti 1988). So, potentially, the population increase at Besnard Lake in the 1970s and 1980s had to do with a reduced persecution and higher winter survival, and may have been the tail end of an earlier increase. Interestingly, the total nesting population size increased until ca. 1988, but the number of successful breeding pairs increased only until 1977, i.e., 10 yr earlier, suggesting two carrying capacities for this population of ca. 26 territorial pairs and of ca. 15 successful breeding pairs, respectively. In parallel with the initial population increase, annual production increased until ca. 1981, potentially contributing to local recruitment and the increase in the number of occupied eagle territories until the late 1980s. However, annual production decreased afterwards, from an average of 30–35 young fledged per year in the 1980s to 20–25 young fledged per year in the 2000s (Fig. 1b). Throughout the whole study period (1968–2011), the proportion of territorial eagles successfully fledging young declined linearly (Fig. 1c). To better understand these changes over time, we looked at density-dependent and climate effects on breeding performance.

We found strong evidence of density-dependence in the Besnard Lake Bald Eagle population. Population growth rate from year to year was strongly negatively related to population density (Fig. 2a), indicating that the population had reached its carrying capacity and was regulated. Bald Eagle breeding performance parameters (annual production, nesting success, mean productivity, and mean brood size at fledging) also negatively correlated with population density (Fig. 2b), indicating that reproduction may be one of the demographic targets of density-dependent regulation.

In raptors, density-dependent reproduction has been found in many species and populations that have been studied. It has been explained in terms of either reduced prey availability (Houston and Schmutz 1995), an increased use of territories of lower quality (Ferrer and Donazar 1996) or a negative impact of floaters (i.e., subadults or adults that are not associated with specific nesting territories and do not reproduce) on the performance of breeding birds (Carrete et al. 2006a, Bretagnolle et al. 2008). Floaters often account for a large proportion of the population at high density (Kenward et al. 2000), and, when abundant, they may strongly affect breeding population properties such as density-dependence (Penteriani et al. 2005, 2006, Bretagnolle et al. 2008). A negative relationship between reproduction and density may also result from habitat heterogeneity, as at low density only the best habitats are occupied (Ferrer and Donazar 1996), or from a buffer effect, with younger or lower-quality individuals occupying lower-quality sites thus reducing average reproductive rates (Ferrer et al. 2006). Sequential habitat occupancy (whereby the best habitats for breeding are occupied first and the poorer habitats occupied subsequently) may therefore be an alternative explanation for a decline in productivity with higher breeding density, as the increase in the number of pairs may have led birds to use poorer habitats. In that case, however, the birds in good habitats should show a constant breeding performance (Mearns and Newton 1988, Ferrer and Donazar 1996). In another study, density-dependent declines in Bald Eagle reproduction were found to be independent of nesting site quality or territory (Elliott et al. 2011). Unfortunately, we had no precise information to control for territory quality, or marked birds of known age to control for difference in breeding experience. Previous works suggest that food supply rather than nest-site availability explains variation in Bald Eagle densities in Saskatchewan (Dzus and Gerrard 1993). Moreover, at Besnard Lake the variance in productivity did not appear to have increased, but rather decreased during the study period (J. Gerard, E. Dzus, and G. Bortolotti, unpubl. data), suggesting that density-dependence similarly affected all pairs. An alternative explanation may be regulation by competition, which may occur through two main processes: direct behavioral interference and resource depletion (Carrete et al. 2006a, Bretagnolle et al. 2008). Although interference refers to food intake rate, it can also apply to territory establishment or mating and mate choice (e.g., Mougeot et al. 2002, Bretagnolle et al. 2008). The higher competition for food resources when the population is at carrying capacity, or a higher interference competition when there are more breeding birds and floaters around, may drive the density-dependent effects on reproduction. Boat surveys of Bald Eagle numbers on Besnard Lake were carried out between 1976 and 2011 (J. Gerrard unpubl. data). The numbers of adults in the first two weeks of July on half the lake slightly increased throughout that time (being 32 in 1976, 32 in 1979, 35 in 1984, 35 in 1990, and 35 in 2003–2005), whereas the number of immature eagles (1 to 3 yr of age) strongly decreased (21 in 1976, 21 in 1979, 10 in 1984, 14 in 1990 and 6 in 2003–2005). The observed decline in mean productivity was not strong enough to explain this decline in the abundance of immature eagles, so these data support the contention that the adult population of Bald Eagles became saturated by the early 1980s and that immature eagles may have been crowded out and forced to forage off Besnard Lake. A crowded population may be more exposed to interference competition, with more intrusions by nonbreeding eagles at nest sites and more competition at foraging sites, which may ultimately result in a poorer breeding performance (Carrete et al. 2006a, Bretagnolle et al. 2008). The observation that the breeding performance of successful pairs was negatively related to the number of failed nesting birds in the population is consistent with this idea, but further work is needed to determine the mechanism of density-dependent depression of reproduction. Data on differences in breeding performance at the territory level, and on the distribution of productivity each year should be used in the future in order to determine the exact mechanisms of density dependent reproduction (Ferrer et al. 2006) but are unfortunately unavailable at this stage for the Besnard Lake population.

Variation in annual breeding performance was mostly explained by density-dependence, much more so than by local climate. We found only a weak positive effect of average spring temperatures on breeding performance, which explained <5% of the variation, whereas density-dependence alone explained 32% of the variation in mean productivity. A positive association between spring temperature and nesting success and mean productivity is consistent with previous studies on Bald Eagles (Gerrard et al. 1992), and adverse weather conditions are known to cause egg failure or the death of young nestlings, thereby causing nest abandonment, as well as reducing food provisioning rates to young (Stalmaster and Plettner 1992, Steidl et al. 1997, Elliott et al. 2005). A weak influence of local climate is not surprising, as the negative effects of climate on demographic rates, and therefore on population growth rates are likely to be greatest at range boundaries (Caughley 1988, Garcia and Arroyo 2001) and Besnard Lake is within the core of the Bald Eagle range. In 1968–2007, local climate change was apparent, in terms of increasing summer temperatures, but this was unlikely to explain the changes in mean productivity or the observed population trends. We had no evidence that local climate influenced population growth rate. However, future studies should look for impacts of climate and density-dependence on wintering areas, which may affect overwinter survival, the condition of Bald Eagles upon their return to their breeding grounds, or the timing of their return, and thereby have indirect effects on their reproduction.

The long-term data on reproduction also showed evidence of regular cyclic fluctuations in Bald Eagle breeding performance (mean productivity, and more specifically, nesting success; Fig. 1c; Fig. 3). After taking into account the effect of density on reproduction, we found that nesting success fluctuated with a regular 5-yr period (Fig. 3). This may be related to fluctuations in food abundance, a primary factor affecting the reproduction of eagles (Whitfield et al. 1974, Whitfield and Gerrard 1985, Gerrard et al. 1992, Anthony et al. 1999; Dykstra et al. 1998, Hoff et al. 2004). Fish are an important part of the diet of Bald Eagles both in breeding and wintering areas (Gerrard and Bortolotti 1988) and fish populations are often characterized by predictable cyclic fluctuations in abundance, with a cycle period equal to the time taken to reach maturity (Kendall et al. 2002, Townsend 2006). Future studies should further investigate which factors drive these fluctuations in Bald Eagle breeding performance, and in particular if some of the main prey species of Bald Eagles at Besnard Lake (i.e., cisco or white sucker) fluctuate in abundance. Factors influencing cyclic variation in eagle productivity may also relate to fluctuating conditions or resources on the wintering areas or areas used during the return migration, with potential carry-over effects on the condition of birds prior to breeding.