SEARCH

SEARCH BY CITATION

Keywords:

  • density dependence;
  • habitat quality;
  • mortality;
  • nesting success;
  • population dynamics

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

1. There are a number of models describing population structure, many of which have the capacity to incorporate spatial habitat effects. One such model is the source–sink model, that describes a system where some habitats have a natality that is higher than mortality (source) and others have a mortality that exceeds natality (sink). A source can be maintained in the absence of migration, whereas a sink will go extinct.

2. However, the interaction between population dynamics and habitat quality is complex, and concerns have been raised about the validity of published empirical studies addressing source–sink dynamics. In particular, some of these studies fail to provide data on survival, a significant component in disentangling a sink from a low quality source. Moreover, failing to account for a density-dependent increase in mortality, or decrease in fecundity, can result in a territory being falsely assigned as a sink, when in fact, this density-dependent suppression only decreases the population size to a lower level, hence indicating a ‘pseudo-sink’.

3. In this study, we investigate a long-term data set for key components of territory-specific demography (mortality and reproduction) and their relationship to habitat characteristics in the territorial, group-living Siberian jay (Perisoreus infaustus). We also assess territory-specific population growth rates (r), to test whether spatial population dynamics are consistent with the ideas of source–sink dynamics.

4. Although average mortality did not differ between sexes, habitat-specific mortality did. Female mortality was higher in older forests, a pattern not observed in males. Male mortality only increased with an increasing amount of open areas. Moreover, reproductive success was higher further away from human settlement, indicating a strong effect of human-associated nest predators.

5. Averaged over all years, 76% of the territories were sources. These territories generally consisted of less open areas, and were located further away from human settlement.

6. The source–sink model provides a tool for modelling demography in distinct habitat patches of different quality, which can aid in identifying key habitats within the landscape, and thus, reduce the risk of implementing unsound management decisions.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The dynamics of natural populations can be influenced by a variety of factors, ranging from density-independent processes such as weather or catastrophic events, to density-dependent factors such as competition over resources or mates (Allee et al. 1949; Barnard 1980; Sutherland 1996). The majority of these factors are affected by characteristics of the habitat in which an individual resides. Despite this, surprisingly many studies of population dynamics do not account for habitat heterogeneity (Pulliam 1996). Fortunately, this trend is turning, and a great amount of current research incorporates spatial aspects in the estimation of population dynamics (Hanski 1999). One of the most established of such spatial models is the source–sink model, in which a source is defined as a habitat where natality exceeds mortality whereas a sink is defined as a habitat in which natality is lower than mortality. If a population is at equilibrium, a source habitat has an emigration that exceeds immigration whereas a sink habitat will show the reverse pattern (Holt 1985; Pulliam 1988; Pulliam & Danielson 1991). A defining feature of source–sink models is that they are deterministic and as such, the differences between sources and sinks are intrinsic; hence, extinction in a sink is inevitable in the absence of immigration (Dias 1996).

Source–sink dynamics are generally applied at a between population level, but the model can also be applied within populations, at the level of the territory. At this scale, a source is a territory where the production of fledglings exceeds breeder mortality in the absence of migration, whereas a sink has a reproduction that is insufficient to replace the breeders (Breininger & Carter 2003). This relationship can be translated into territory-specific population growth rates, where a source will have a positive population growth rate (r > 0 or λ > 1), and a sink will display a negative population growth rate (r < 0 or λ < 1). The features of source–sink dynamics make them particularly suitable for modelling demography in distinct habitat patches of different quality and, hence, source–sink dynamics have gained considerable popularity in conservation biology (Harrison & Bruna 1999). The large interest in the model has generated many studies that claim to provide empirical support for source–sink dynamics (Pulliam 1996). However, the majority of these studies lack sufficient data to separate a source and a sink from simply a good or a bad quality habitat (Diffendorfer 1998; Runge, Runge & Nichols 2006). In particular, there is a widespread lack of survival estimates, which are vital in deciding whether a habitat is a true sink or just a low quality territory. Moreover, many studies have failed to assess density dependence hence precluding a distinction between a ‘true sink’ and a ‘pseudo-sink’ (Watkinson & Sutherland 1995). A ‘pseudo-sink’ arises when density-dependent immigration into a source causes an increase in total mortality or a depression in fecundity. Thus, in the absence of migration, a population in a pseudo-sink will only decline to a lower level, whereas in a true sink, the population will go extinct (Watkinson & Sutherland 1995; Pulliam 1996).

In this study, we use long-term population data to investigate key components of population demography (nesting success, offspring production, and adult mortality) and assess the spatial population dynamics at the level of the territory, in a sedentary population of Siberian jay (Perisoreus infaustus). We also investigate environmental factors that are likely to affect the different components of population demography and spatial dynamics. Our study population is well suited for the study of habitat-specific demography and spatial dynamics in many respects. First, annual data on survival and reproductive success in this population have been collected since 1990. This long-term data will reduce the risk of any observed demographic pattern being an artefact of temporary effects. Secondly, annual estimates of population sizes and territory-specific group sizes allow us to assess density-dependent effects on demographic parameters. Thirdly, the study site contains a range of different habitats, largely due to changes induced by humans. In particular, extensive forest management has created forests of different openness and structure. Previous studies on this population have found a pronounced effect of forest structure on reproductive success (Eggers et al. 2005; Griesser et al. 2007) and juvenile mortality (Griesser, Nystrand & Ekman 2006), demonstrating that an open forest containing less understorey provides a lower quality habitat. Moreover, predation is the main cause of reproductive failure (Eggers et al. 2005) and mortality in this population (Griesser et al. 2006). The main nest predators are different species of corvids, and the main predator on adults and juveniles is the goshawk, Accipiter gentilis (Eggers et al. 2005; Griesser et al. 2006). All these predators use visual cues to locate their prey, and thus, an open and fragmented forest facilitates hunting success in these species (Griesser et al. 2007). Hence, this study system not only allows us to efficiently address aspects of population dynamics per se, but it also facilitates a deeper understanding of these demographic patterns by enabling us to harness a broad existing data base on the species’ ecology.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study site and species

The study was conducted between 1990 and 2004 on a population of individually colour-banded Siberian jays near Arvidsjaur, northern Sweden (65°40′ N, 19°0′ E). Individual territories (N = 21) were located within different forest structures, ranging from forest plantations (age < 50 years) and heavily managed forest consisting mainly of young pine, (Pinus sylvestris), to more diverse forest stands consisting of a mixture of spruce (Picea abies), pine, birch (Betula pubescens) and aspen (Populus tremula). The managed forests have been subject to commercial forestry, which involves thinning, harvesting and re-planting, a cycle that takes about 80–120 years in this area (Loman 2005). Thinning involves removing most of the understorey in the forest (after around 25, 40 and 60–80 years), whereas re-planting usually involves the plantation of pine trees. Territories within the study site were contiguous with the exception of four territories that bordered unsuitable habitat.

Siberian jays live in social groups of two to seven individuals that, in addition to the breeding pair, may contain retained offspring and/or unrelated immigrants (Ekman, Sklepkovych & Tegelström 1994). The breeding pair forms life-long relationships, and there are no recorded cases of extra-pair paternity (H. Tegelström, pers. comm., 2001; Lillandt, Bensch & von Schantz 2001). They produce one brood per season with clutch sizes ranging from 1 to 5 (average 3·9 ± 0·1; Eggers et al. 2006). Approximately one-third of all fledglings delay dispersal and remain in their natal territories for up to 3 years (Ekman et al. 2001; Griesser et al. 2008). The fledglings that remain at home are the dominant brood members, and they actively expel the subordinate brood mates from the parental territory 3–6 weeks after fledging. These expelled individuals disperse and immigrate into other groups (Griesser et al. 2008). Breeders are nepotistic towards their retained offspring, providing them with access to resources and predator protection, benefits that are withheld from immigrant group members (Ekman et al. 1994; Griesser 2003, 2008; Griesser & Ekman 2004, 2005). Neither retained offspring, nor immigrants, engage in allofeeding (i.e. bringing food to others; Ekman et al. 1994), participate in anti-predator activities (Griesser 2008), or augment reproductive success or survival of the breeding pair (Ekman et al. 1994; Griesser et al. 2006). Furthermore, retained offspring rarely inherit the natal territory at the death of a parent (Ekman et al. 2001; Ekman & Griesser 2002; Kokko & Ekman 2002), probably a result of inbreeding avoidance or reproductive conflicts (Griesser et al. 2008). Rather, breeding vacancies are quickly filled by non-related immigrants (Ekman & Griesser 2002). Furthermore, there are no records of a breeder being evicted from a territory by a kin or an immigrant, and breeders rarely divorce their partners and disperse (n = 5 out of 220 recorded pairs; J. Ekman, unpublished data). Thus, territory-acquisition almost exclusively occurs following the death of a breeder.

Data collection of species data

We only included territories where we had data on both reproductive success and survival. Nine of these territories were vacated as a consequence of forestry during the course of the study, at which time calculations were terminated (mean number of years studied 5·1 ± 2·5 SD, range 3–10 years). Two territories were severely altered by forestry which resulted in them being vacated. They were later re-occupied by new breeders and thus treated as different territories in the analyses.

Territories were visited at least twice per year to record group composition and reproductive success. Prior to breeding, the female breeder was fitted with a radiotransmitter (Holohil, Canada, weight = 1·76–1·85 g, ≈ 2·0% of the body weight), and monitored until the nest was found. Nestlings were banded in the nest (n = 339) in May. In the cases where the nest could not be located, individual offspring were banded after fledging (usually within 5 months, n = 73; but in three cases offspring were banded in their second year). Immigrants were mostly banded during summer and autumn (n = 224). In 146 cases, origin of non-breeders was unknown. In these cases, we assessed their relatedness to the breeders using behavioural interactions. This method has proven reliable, both from comparison with individuals of known origin (Griesser 2003) and by verification using genetic fingerprinting (Ekman et al. 1994). The sex of each bird was determined using the molecular technique described by Griffith et al. (1998).

Data collection of environmental parameters

Environmental parameters were extracted from vegetation maps and aerial photographs, using ArcGis 9.1 (Geographical Information System; ArcView/Arc GIS, GIS & mapping software, Redlands, USA). We digitally applied a 30-ha circular buffer around the breeding area which allowed us to focus solely on the ‘nuclear’ territory. The size of this buffer was based on previous data (from several years) that indicated an average territory size of 66 ha (range: 41–108 ha, n = 12 territories; M. Nystrand, unpublished data). The use of a buffer was motivated by insufficient data on territory borders. Siberian jays do not have strict outer territory borders, but rather, they seem to concentrate the majority of time to a core area around the breeding site. Thus, the buffer should adequately capture essential environmental characteristics of the territory.

The environmental parameters were selected based on their expected influence on mortality risk and reproductive success, either by the way they illustrated the structure of the forest or by an expected relationship to predator abundance. Specifically, the selected parameters were volume of spruce and pine, forest age, tree height, area of natural openings (e.g. mosses, meadows), distance to the nearest human settlement, and the number of roads intersecting the territory. After investigating patterns of multicollinearity, only four parameters were suitable for further analysis: average age of the forest (years), average volume spruce (m3 ha−1; only including forested areas), area of natural openings (m2), and distance to nearest human settlement (m; measured from the central breeding area). The age of the forest is related to the structure of the forest, where a repeated thinning of the forest has resulted in an increasingly open forest structure. In the study area, the forest was never more than about 80–100 years old, suggesting that the territories containing oldest forests also had a more open forest structure. The unit describing volume reflects how ‘dense’ the forest is, whereas the area of natural openings is likely to illustrate the presence of edge zones (hence, fragmentation) in the territory. Finally, the distance to nearest human settlement has previously been found to affect the risk of nest predation (Eggers et al. 2005), most likely because other corvids (e.g. common jay, crows) aggregate around human settlement because of an increased access to food (e.g. bird feeders).

Data analysis

Because of a continuous expansion of the study site over the years (Table 1), we tested if the overall population was increasing or decreasing over years by analysing changes in average group sizes between years rather than total population size, using a linear regression. Areas surrounding the study site have frequently been sampled for dispersers, and average dispersal distances have been recorded to be no more than 2·3 territories away from the natal territory (Ekman, Eggers & Griesser 2002; Kokko & Ekman 2002). Also, detection probabilities demonstrate that the number of undetected individuals is negligible (<1%: Ekman, Bylin & Tegelström 2000; Ekman et al. 2001). Thus, despite a lack of sufficient data on dispersal out of the study site, it is unlikely that a surplus production of young resulting in a large-scale emigration out of the area would go undetected. Rather, we expect that a substantial increase in population size will, at least partly, result in an increase in group sizes within the area.

Table 1.   Population sizes in different subsamples. Estimates of group and population sizes consist of values averaged over the focal year, since group composition may fluctuate within a given territory and year. Only occupied territories are included in these values. Total average bird density per territory is measured over the same territories and years, thus including both occupied and abandoned territories
YearNumber of territoriesAverage bird density per territory Average population sizeaTotal average bird density per territory (n = 15 territories)
  1. aBecause of an increased sampling effort, the average population size increases with year. Hence, this increase does not describe an increase in population size per se.

199052·211·0
1991102·622·0
1992133·037·02·6
199383·528·02·7
1994123·740·52·9
199593·330·02·9
199692·218·01·7
1997122·124·02·1
199893·027·02·3
1999153·450·52·6
2000112·931·53·4
2001172·949·02·1
2002222·758·51·7
2003183·360·01·2

Data on adult breeder mortality (henceforth referred to only as ‘female or male mortality’), nesting success and offspring production was extracted for each territory. Only data from territories that had not been subject to prior experiments (e.g. supplementary feeding, removal of breeders) that could have affected any of these factors was included. Adult mortality was measured by monitoring the group composition of each territory each year, assuming that breeders that had disappeared were dead. The high detection probabilities (see above), and the fact that breeders rarely change territories (Griesser et al. 2007), support this assumption. Nesting success was measured as the probability of producing at least one successful fledgling (failure or success), whereas offspring production was estimated as annual fledgling productivity.

We tested if breeder mortality differed according to sex, group size and year (class variable), using a Generalized Linear Mixed Model (GLMM) with binary error distribution and logit link. Territories were fitted as random effects, and ‘time’ (with territory ID as the subject) as a repeated effect to control for repeated observations from within the same territory being more dependent when closer in time. A similar statistical model was used to test for the environmental influence on sex-specific mortality and nesting success, where the independent variables were year, average age of the forest, average volume of spruce, total area of natural openings and distance to nearest human settlement. We also tested for density-dependent mortality by analysing whether population size had a negative effect on breeder mortality, using a GLMM (binary distribution and logit link). A similar test was used to investigate if there was a territory-specific density-effect of general population size on fledgling production, and of spring group size (i.e. breeding season) on fledgling production (GLMM, Poisson error and logarithm link). A negative relationship with population size would indicate that a larger population suppressed the production of young. The data in the study were slightly unbalanced because values of the number of fledglings and environmental parameters were not available for all territories in all years.

Territory-specific dynamics

Spatial population dynamics was measured by assessing annual territory-specific population growth rates (r), where the estimations of r were based on a discrete growth model. Specifically, annual birth and death rates were calculated for each territory. From these values, we estimated λ (i.e. 1 + − d), which was later transformed (thus achieving a normal distribution) to r (ln λ). Population growth rates (r) were tested against the selected environmental parameters. We also tested if there was a difference in the number of related individuals and immigrants between the territory-specific classes of sources (r > 0) and sinks (r < 0) (averaged over years because many years displayed a stable population growth, thus not strictly confirming to a source or a sink; GLMM, normal error, identity link). We expected source territories to have a higher fledgling production than a sink territory. In addition, a source may also contain more offspring that have delayed dispersal and remained in the natal territory because of the beneficial conditions associated with home (Griesser et al. 2006). Thus, in a social system such as that of the Siberian jay, a sink might not contain more immigrants than a source simply because offspring from high producing territories temporarily forego breeding and remain at home, rather than disperse into a poor quality territory. Hence, in this system, a source should contain more related birds than a sink, and less, or equally many, immigrants compared to a sink.

Correlating factors

A linear regression was used to test if the distance to human settlement was related to nest predator abundance, a relationship that has been found in previous studies (Ekman et al. 2001; Eggers et al. 2005). Data of corvid activity was collected during the breeding season around nest sites by counting the number of corvids (i.e. Eurasian jay Garrulus glandarius, hooded crow, Corvus corone and raven Corvus corax) for 30 min in 5-min intervals on daily, randomly timed, visits to the nest sites (for more details, see Eggers et al. 2005; Eggers, Griesser & Ekman 2008).

General statistics

All statistical analyses in this study were performed in sas v. 9.1 (SAS Institute Inc., Cary, NC, USA). Models were checked for violations against model assumptions and models containing outliers were run both with the full data set and with the outliers deleted to investigate how they affected the results. We selected the final model by including the subset of parameters that generated the best adjusted r2 (Quinn & Keough 2002), or Akaike Information Criteria, (AIC or AICc; for the GLMMs). In the cases where only pseudo-AIC were given (i.e. GLMMs with non-normal distribution), we assessed the residuals of each model and compared the random model (producing pseudo-likelihood values, which cannot be used to estimate model fit; The glimmix procedure, June 2006. SAS/Stat (R) 9.2 User’s guide, http://www.support.sas.com) to the equivalent fixed effect model (that did not contain the random effects, hence producing maximum likelihood values which generate comparable AIC values).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Population dynamics

The average number of birds per territory (density) showed no consistent trend over the study period (Table 1; linear regression, r2 = 0·02, F1,12 = 0·20, P = 0·66). However, this analysis only included territories that were occupied and that had not been subject to intense forestry. In comparison, including empty and/or severely altered territories in the analysis over population growth (fixed number of 15 territories, ranging over 12 years) demonstrated a significant decrease in population size (Table 1; linear regression, r2 = 0·50, F1,11 = 10·01, P = 0·010).

Breeder mortality

Average breeder mortality in the entire population and over the whole study period was 0·24 ± 0·03 (absolute mean ± SE; 0·25 ± 0·04, n = 106 individuals, for females and males respectively). Breeder mortality was not related to population size (GLMM, binary error, logit link: n = 230, F1,129 = 0·12, P = 0·73). Moreover, there was no difference in mortality between sexes (GLMM, binary error, logit link: n = 212, F1,90 = 0·24, P = 0·63), year (F13,106 = 0·93, P = 0·53) or group sizes (F1,106 = 0·51, P = 0·48).

Correlation with environmental parameters

Female and male mortality showed different relationships to environmental parameters. Female mortality was positively related to the age of the forest (Table 2, Fig. 1a), but was not affected by any of the other parameters. Male mortality increased with larger areas of natural openings in the territory (Table 2, Fig. 1b), but showed no relationship to any other variables.

Table 2.   Effects of environmental parameters on mortality in females and males (n = 19 territories, respectively, GLMM, binary error, logit link). Emboldened text denotes the best fitting model according to chosen model selection method (see Methods for more information). Normal text denotes results of the full model
 FemalesMales
Num. d.f.Den. d.f.FP-valueNum. d.f.Den. d.f.FP-value
Year13740·490·924713740·750·7092
Average age of forest1874·150·04471871·190·2775
Average volume spruce1872·610·11011740·060·8097
Area natural openings1742·120·14951874·710·0327
Distance settlement1740·690·40831740·390·5336
image

Figure 1.  (a) Average female mortality (±SE) in relation to the age of the forest (nterritories = 19). The displayed values are averages of the binary values (i.e. alive or dead) over all years for each territory. Also displayed is the number of years included in the calculations of the average values. (b) Average male mortality (±SE) in relation to the area of the territory consisting of natural openings (nterritories = 19). The displayed values are averages of the binary values (i.e. alive or dead) over all years for each territory. Also displayed is the number of years included in the calculations of the average values.

Download figure to PowerPoint

Reproductive success

Average nesting success in the population over all years and territories was 0·44 ± 0·04 (absolute means ± SE, n = 115 nests) and the associated average fledgling output was 1·37 ± 0·15 (n = 117). The average fledgling production was similar over years, with no time-trend in any direction (linear regression, r2 = 0·32, F2,11 = 2·53, P = 0·12). Moreover, the number of fledglings in the population was not related to population size (GLMM, Poisson error, logarithm link: n = 115, F1,93 = 1·69, P = 0·20), suggesting that there was no density-dependent effect on reproductive success. Likewise, fledgling production was unrelated to spring group size (GLMM, Poisson error and logarithm link: n = 114, F6,74 = 0·50, P = 0·80), which further confirmed a lack of density-dependent suppression of reproduction.

Correlation with environmental parameters

Nesting success was not associated with the average age of the forest, volume of spruce, year, or area of natural openings. However, nesting success did increase with an increased distance to human settlement (Table 3).

Table 3.   Effects of environmental parameters on nesting success (n = 17 territories, GLMM, binary error, logit link). Emboldened text denotes the best fitting model according to chosen model selection method (see Methods for more information)
 Num. d.f.Den. d.f.FP-value
Year13691·020·4436
Average age of forest1690·120·7340
Average volume spruce1811·100·2983
Area natural openings1811·100·3921
Distance settlement1816·080·0158

Territory-specific dynamics

Out of a total of 21 territories, 16 had an overall positive population growth (r), suggesting that they were source territories. As expected, the number of retained offspring was higher in the average source than in the average sink (GLMM, normal error distribution, identity link: mean source ± SE = 0·57 ± 0·10 vs. sink = 0·05 ± 0·18, n = 21, F1,19 = 6·37, P = 0·020). However, the number of immigrants did not significantly differ between sources and sinks (GLMM, normal error distribution, identity link: mean source ± SE = 0·58 ± 0·11 vs. sink = 0·49 ± 0·20, n = 21, F1,19 = 0·15, P = 0·70), further supporting the conclusion that the designated sinks were true sinks and not pseudo-sinks.

Territory-specific population growth decreased with an increase in natural openings in the territory (opposite pattern to Fig. 1b: GLMM, normal error distribution, identity link: F1,32·7 = 5·56, P = 0·025). Moreover, there was a trend (close to significant) towards higher territory-specific population growth rates further away from human settlement (Fig. 2: F1,43·6 = 3·52, P = 0·067). Distance to human settlement was strongly related to the abundance of nest predators, demonstrating that territories closer to human settlement had a higher abundance of nest predators than territories further away (linear regression, r2 = 0·35, F1,16 = 8·55, P = 0·01). Finally, there was a strong effect of year (Table 4; F13,58·5 = 2·25, P = 0·018) on population growth rates.

image

Figure 2.  Territory-specific growth rates (±SE) in relation to the distance to human settlement (n = 17). An average r-value above 0 indicates that a territory has been successful (i.e. a source) or stable most or all years of sampling, whereas a value below 0 indicates that a territory has repeatedly failed.

Download figure to PowerPoint

Table 4.   Yearly fluctuations in population growth rates for each territory. Territories below the hatched line have an overall negative population growth (i.e. sinks)
TerritorySource (r > 0)Sink (r < 0)Stable (r = 0)
A103
B312
C210
D550
E410
F210
G403
H432
I301
J111
K210
L212
M404
N334
O300
P300
Q111
R142
S021
T022
U044
V113

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The source–sink model is conceptually well-developed. However, many of the empirical studies aspiring to demonstrate source–sink dynamics have failed to meet the basic assumptions associated with the model (reviewed in Diffendorfer 1998, and Runge et al. 2006; but see Breininger and Carter 2003,Breininger & Oddy 2004). In this study, we applied the theory of source–sink dynamics to a bird population, in which we had sufficient demographic data to meet the requirements of the model. The results indicate that our system is a true source–sink system, but that the population dynamics of individual territories fluctuates between years.

Population dynamics and individual demographic factors (mortality and reproduction)

The size of the study population did not show a consistent trend of increase or decrease over the years of study. When including territories that had been subject to intense forestry (e.g. extensive logging or thinning) to the analysis, the population demonstrated a negative growth. Negative effects of habitat loss and degradation can sometimes affect untargeted neighbouring territories (i.e. edge effect; Wiens 1997). We found no indications of such a ‘spill-over effect’ in our population, as indicated by the non-significant change in the number of birds per monitored territory. However, altered territories have at times been found to have a delayed response to forestry (Griesser et al. 2007). Hence, it is possible that we will see a deferred decrease in population sizes in the future, which is the result of past habitat alterations.

Breeder mortality

There was no difference between sexes in overall mortality rates. However, there was a clear difference between sexes in relation to environmental factors. Female mortality was affected by the age of the forest (higher mortality in older forests), whereas male mortality was affected by the amount of open areas in the territory (higher mortality in more open areas). This discrepancy could be a result of sex-biased environmental sensitivity (Råberg, Stjernman & Nilsson 2005), implying different mechanisms underlying sex-specific patterns of mortality in this species. Siberian jay females appear to be more sensitive to an overall more transparent, less dense habitat (older forests are more open), which is related to higher predation risks because it increases the hunting success for predators of jays (not because of differences in predator abundance since it is similar throughout the area; Griesser et al. 2006). We contemplate that this potentially higher vulnerability among females to predation may be related to reproductive costs (Liker & Székely 2005; Owens & Bennett 1994); because females carry additional reserves and eggs at the onset of breeding (J. Ekman, unpublished data), they are less likely to succeed in escaping a predator attack at this time.

In contrast, male mortality appears to be related to general risk-taking behaviour. Siberian jay males spend a vast amount of time feeding in open areas, in particular during the breeding season when they have to provide for themselves and the female (M. Nystrand, unpublished data). Based on the hunting behaviour of the Siberian jay’s main predator, the goshawk, these areas are likely to be associated with higher risks. Goshawks show a strong preference for woodland edges (0–200 m from open habitat) and the majority of successful kills are made within this zone (Kenward 1982). Previous behavioural observations of Siberian jays demonstrate that open areas are perceived as more risky (Nystrand 2006, 2007; Griesser & Nystrand 2009), and that males take greater risks than females when confronted with a predator (Griesser 2003; Griesser & Ekman 2004, 2005).

Nesting success

The average probability of a nest to produce at least one offspring was only 0·44. Nest predation by corvids is intense and it is the primary cause of nest failure in this population (Eggers et al. 2005). Many corvids benefit from human-altered habitats (Andrén 1992), which allow them to reach higher abundances than they otherwise would. The positive relationship between corvid abundance and the distance to human settlement in this study indicates such an association in our study site. Thus, Siberian jays respond to altered habitat-conditions by preferring territories further away from human settlement, as indicated by the choice of territories made by high quality individuals (Ekman et al. 2001; Eggers et al. 2005).

Our data showed no effect of forest structure per se on nesting success, a result which deviates from previous studies on this species (Ekman et al. 2001; Eggers et al. 2005). This discrepancy is likely to be related to the scale of measurement. Here, forest structure was measured on 30-ha plots, using remote sensing procedures. Previously, forest structure has been measured in the field, in the immediate surroundings of nest sites (0·02–0·3 ha). Thus, forest cover at the smaller scale appears to be more important in concealing the nest than the large-scale territory habitat.

Identifying spatial dynamics – a combination of demographic factors

We assessed territory quality based on territory-specific growth rates. The subsample of territories in this study only included territories that were unaffected by forestry during the period of sampling, during which the overall population growth did not decrease. Hence, we expected most territories to be sources. When averaging the demographic values over years for each territory, 16 territories fulfilled the assumptions of a source, representing 76% of all territories. The overall growth rates varied between years however, with year 1996 and 2000 displaying lower growth rates than most other years. Also, focal territories appeared to fluctuate randomly between being a source and a sink, and were frequently interspersed with years during which there was no change in population growth (i.e. stable, r = 0). This suggests that our population confirms to a temporally varying source–sink system, where potentially small changes in reproductive success or mortality between years may be enough to change the state of a territory. However, without more long-term data, it is not possible to identify potential time-trends in these fluctuations.

A larger area of natural openings within a territory had a negative effect on territorial growth rates. This variable was previously found to influence male mortality, suggesting that male survival is crucial for the overall success of a territory. This is likely to be a result of his important role in supplying food to the female and the nestlings during nesting. The other environmental variable that affected (approaching significance) territory dynamics was the distance to human settlement, with increasing distances resulting in higher growth rates. This was also the sole factor influencing nesting success. Hence, these results combined suggest that it is mainly the performance during the actual breeding event that determines the overall state of a territory. Failure (e.g. female mortality) during the remaining year may be easier to compensate for, and will thus have less impact on population dynamics, than failure that ensues after the onset of breeding.

Many studies assessing source–sink dynamics in natural systems have been criticized for not properly accounting for factors that might confound demographic estimates (e.g. by neglecting density dependence or not accounting for dispersal (Dias 1996; Diffendorfer 1998; Watkinson & Sutherland 1995). For example, failing to account for density dependence can, despite sufficient demographic data, falsely classify a source as a sink (a so called ‘pseudo-sink’; Watkinson & Sutherland 1995). It is unlikely that we have falsely classified sources as sinks in this study. There was no effect of population size or group size on fledgling production or on breeder mortality, indicating a lack of a density-dependent suppression on reproductive success and mortality. In addition, we did not find any difference in the number of immigrants between sources and sinks, which indirectly supports a lack of density-dependent (i.e. interference caused by immigrants). Our findings are congruent with previous studies that found no evidence that non-breeding group members augment reproductive success (Ekman et al. 1994) or decrease breeder survival (Griesser et al. 2006, 2008). Furthermore, there is no evidence that interference with neighbours affects reproduction or mortality and territory owners are never displaced once they have gained a territory (Ekman et al. 2001).

The high re-sighting score (>99%) of individuals in our study population (Ekman et al. 2000, 2001) makes it highly unlikely that a significant number of individuals would have escaped our censuses. Thus, the lack of dispersal data in this study is unlikely to undermine the interpretation of the results. In fact, since about 50% of Siberian jay offspring delay dispersal for up to 3 years (Ekman et al. 2001), failing to acquire consistent long-term data could falsely indicate that a territory had low or no emigration (thus being a sink), when in reality, the low emigration rate is a simple effect of offspring delaying dispersal for a maximum time because the territory is of high quality (hence, being a source).

Conservation implications

There can be substantial variation between different groups of individuals in their response to habitat degradation, depending on factors such as sex, mode of territory acquisition or territory choice. Hence, simply estimating population growth by numbers can give misleading results (Pulliam 1988, 1996; Kokko, Sutherland & Johnstone 2001). The results of this study emphasize the significance of both choosing an adequate scale of sampling, and selecting the appropriate demographic parameters in the assessment of population dynamics. Furthermore, our data demonstrates that territories can vary temporally between being a source and a sink, which illustrates the significance of long-term data in the identification of high quality areas. By applying a source–sink framework to long-term data on population dynamics, the risk of implementing unsound management decisions based on estimates of populations from homogenous habitats or on populations that are unable to maintain themselves in the absence of immigration can be avoided.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We are grateful to Damian Dowling, Jacob Höglund, Derek M. Johnson, and anonymous referees for constructive comments on different versions of the manuscript. We thank Folke & May Lindgren for sharing their knowledge of the jays and Gunnar & Ingrid Pavval for providing a good base at Lappugglan. Also, many thanks to everyone who has helped in the field and laboratory over the years. We thank Tina Granqvist Pahlén, Mats Nilsson and Sveaskog for kindly providing forest maps of the area. This study was supported by the Swedish Research Council (JE), The Swedish Research Council for Environment, Agricultural Sciences, and Spatial Planning (FORMAS) and Swedish Environmental Protection Agency (Naturvårdsverket) (JE). Stiftelsen för Zoologisk forskning (MG), C. F. Lilljewalchs resestipendium (MG), Stiftelsen Alvins fond för fågelskydd (MG, MN), Hiertas Minnesfonds (MG), Helge Ax:son Johnson Stiftlesen (MN) and the Swiss National Science Foundation (MG).

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Allee, W.C., Emerson, A.E., Park, O., Park, T. & Schmidt, K.P. (1949) Principles of Animal Ecology. Saunders, Philadelphia.
  • Andrén, H. (1992) Corvid density and nest predation in relation to forest fragmentation: a landscape perspective. Ecology, 73, 794804.
  • Barnard, C.J. (1980) Flock feeding and time budgets in the house sparrow (Passer domesticus L.). Animal Behaviour, 28, 295309.
  • Breininger, D.E. & Oddy, D.M. (2004) Do habitat potential, population density, and forest influence scrub-jay source-sink dynamics? Ecological Applications, 14, 10791089.
  • Breininger, D.E. & Carter, G.M. (2003) Territory quality transitions and source-sink dynamics in a florida scrub-jay population? Ecological Applications, 13, 516529.
  • Dias, P.C. (1996) Sources and sinks in population biology. Trends in Ecology and Evolution, 11, 326330.
  • Diffendorfer, J.E. (1998) Testing models of source-sink dynamics and balanced dispersal. Oikos, 81, 417433.
  • Eggers, S., Griesser, M., Andersson, T. & Ekman, J. (2005) Nest predation and habitat change interact to influence Siberian jay numbers. Oikos, 111, 150158.
  • Eggers, S., Griesser, M., Nystrand, M. & Ekman, E. (2006) Predation risk induces changes in nest site selection and clutch size in the Siberian jay. Proceedings of the Royal Society of London, Series B: Biological Sciences, 273, 701706.
  • Eggers, S., Griesser, M. & Ekman, E. (2008) Predator-induced reductions in nest visitation rates are modified by forest cover and food availability. Behavioral Ecology, 19, 10561062.
  • Ekman, J. & Griesser, M. (2002) Why offspring delay dispersal: experimental evidence for a role of parental tolerance. Proceedings of the Royal Society of London, Series B: Biological Sciences, 269, 17091713.
  • Ekman, J., Sklepkovych, B. & Tegelström, H. (1994) Offspring retention in the Siberian jay (Perisoreus infaustus): the prolonged brood care hypothesis. Behavioral Ecology, 5, 245253.
  • Ekman, J., Bylin, A. & Tegelström, H. (2000) Parental nepotism enhances survival of retained offspring in the Siberian jay. Behavioral Ecology, 11, 416420.
  • Ekman, J., Eggers, S., Griesser, M. & Tegelström, H. (2001) Queuing for preferred territories: delayed dispersal of Siberian jays. Journal of Animal Ecology, 70, 317324.
  • Ekman, J., Eggers, S. & Griesser, M. (2002) Fighting to stay: the role of sibling rivalry for delayed dispersal. Animal Behaviour, 64, 453459.
  • Griesser, M. (2003) Nepotistic vigilance behaviour in Siberian jay parents. Behavioral Ecology, 14, 246250.
  • Griesser, M. (2008) Referential calls signal predator behavior in a group-living bird species. Current Biology, 18, 6973.
  • Griesser, M. & Ekman, J. (2004) Nepotistic alarm calling in the Siberian jay (Perisoreus infaustus). Animal Behaviour, 67, 933939.
  • Griesser, M. & Ekman, J. (2005) Nepotistic mobbing behaviour in the Siberian jay, Perisoreus infaustus. Animal Behaviour, 69, 345352.
  • Griesser, M. & Nystrand, M. (2009) Vigilance and predation of a forest-living bird species depend on large-scale habitat structure. Behavioral Ecology, 20, 709715.
  • Griesser, M., Nystrand, M. & Ekman, J. (2006) Reduced mortality selects for family cohesion in a social species. Proceedings of the Royal Society of London, Series B: Biological Sciences, 273, 18811886.
  • Griesser, M., Nystrand, M., Eggers, S. & Ekman, J. (2007) Impact of forestry practices on fitness correlates and population productivity in an open-nesting bird species. Conservation Biology, 21, 767774.
  • Griesser, M., Nystrand, M., Eggers, S. & Ekman, J. (2008) Social constraints limit dispersal and settlement decisions in a group-living bird species. Behavioral Ecology, 19, 317324.
  • Griffith, R., Double, M.C., Orr, K. & Dawson, R.J.G. (1998) A DNA test to sex most birds. Molecular Ecology, 7, 10711075.
  • Hanski, I. (1999) Metapopulation Ecology. Oxford University Press, Oxford.
  • Harrison, S. & Bruna, E. (1999) Habitat conservation and large-scale conservation: what do we know for sure? Ecography, 22, 225232.
  • Holt, R.D. (1985) Population dynamics in two-patch environments: some anomalous consequences of an optimal habitat distribution. Theoretical Population Biology, 28, 181208.
  • Kenward, R.E. (1982) Gowhawk hunting behaviour, and range size as a function of food and habitat availability. Journal of Animal Ecology, 51, 6980.
  • Kokko, H. & Ekman, J. (2002) Delayed dispersal as a route to breeding: territorial inheritance, safe havens, and ecological constraints. The American Naturalist, 160, 468484.
  • Kokko, H., Sutherland, W.J. & Johnstone, R.A. (2001) The logic of territory choice: Implications for conservation and source-sink dynamics. The American Naturalist, 157, 459463.
  • Liker, A. & Székely, T. (2005) Mortality costs of sexual selection and parental care in natural populations of birds. Evolution, 59, 890897.
  • Lillandt, B.-G., Bensch, S. & Von Schantz, T. (2001) Parentage determination in kin-structured populations: microsatellite analysis in the Siberian jay Perisoreus infaustus during a 25-year population study. Avian Science, 1, 314.
  • Loman, J.-O. (2005) Skogsstatistisk årsbok 2005. National Board of Forestry, Jönköping.
  • Nystrand, M. (2006) Influence of age, kinship and large-scale habitat quality on local foraging choices of Siberian jays. Behavioral Ecology, 17, 503509.
  • Nystrand, M. (2007) Associating with kin affects the trade-off between energy intake and exposure to predators in a social bird species. Animal Behaviour, 74, 497506.
  • Owens, I.P.F. & Bennett, P.M. (1994) Mortality costs of parental care and sexual dimorphism in birds. Proceedings of the Royal Society of London, Series B: Biological Sciences, 257, 18.
  • Pulliam, H.R. (1988) Sources, sinks, and population regulation. The American Naturalist, 132, 652661.
  • Pulliam, H.R. (1996) Sources and sinks: empirical evidence and population consequences. Population Dynamics in Ecological Space and Time (eds O.E.Rhodes, R.K.Chesser & M.H.Smith), pp. 4569. University of Chicago Press, Chicago.
  • Pulliam, H.R. & Danielson, B.J. (1991) Sources, sinks, and habitat selection: a landscape perspective on population dynamics. The American Naturalist, 137(Suppl.), 5066.
  • Quinn, G.P. & Keough, M.J. (2002) Experimental Design and Data Analysis for Biologists. University Press, Cambridge.
  • Råberg, L., Stjernman, M. & Nilsson, J.-Å. (2005) Sex and environmental sensitivity in blue tit nestlings. Oecologia, 145, 496503.
  • Runge, J.P., Runge, M.C. & Nichols, J.D. (2006) The role of local populations within a landscape context: defining and classifying sources and sinks. The American Naturalist, 167, 925938.
  • Sutherland, W.J. (1996) From Individual Behaviour to Population Ecology. Oxford University Press Inc., New York.
  • Watkinson, A.R. & Sutherland, W.J. (1995) Sources, sinks and pseudo-sinks. Journal of Animal Ecology, 64, 126130.
  • Wiens, J.A. (1997) Metapopulation dynamics and landscape ecology. Metapopulation Biology (eds I.A.Hanski & M.E.Gilpin), pp. 4362. Academic Press, San Diego.