• brown-headed cowbird;
  • GIS;
  • song sparrow;
  • source–sink dynamics;
  • species habitat modelling;
  • refuges;
  • parasitism


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    Conversion of natural habitats to human use can affect the abundance and distribution of predators and parasites, create population sinks, and reduce the viability of valued prey and host species. We asked how the distribution of brown-headed cowbirds Molothrus ater, a generalist brood parasite, has influenced the source–sink dynamics of song sparrow Melospiza melodia populations and their regional population trends.
  • 2
    We intensively studied 17 host populations subject to varying levels of parasitism for 1–36 years. We linked these data to spatial and demographic models to predict growth rate in song sparrow populations in the Southern Gulf Islands, BC, Canada.
  • 3
    Patterns of growth in song sparrow populations were closely related to cowbird distribution, which in turn depended on land use patterns at landscape scales. Locally, sparrow populations were expected to increase in areas far from cowbird feeding areas, where parasitism was low, but to decline where parasitism exceeded 20%. The predicted population trends were similar to those recorded locally and via the North American Breeding Bird Survey.
  • 4
    Synthesis and applications. We show that the distribution of habitats favourable to brood parasites can affect whether host populations grow or decline regionally. In highly sedentary hosts like the sparrows in this study, density-dependent juvenile dispersal and marked spatial variation in the probability of parasitism can give rise to source-and-sink dynamics. Our results illustrate how the application of spatial models and empirical data can predict how land use decisions may influence host dynamics. We identify ways in which applied ecologists might influence land use to enhance the persistence of valued hosts, and suggest that our approach provides a promising framework for exploring regional-scale spatial dynamics of species in order to identify critical habitat and prioritize investments in conservation.


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

The occurrence of source–sink dynamics, wherein sink populations rely on immigrants from more productive sources for their persistence, can arise as a consequence of habitat-specific variation in survival and fecundity (Pulliam 1988) and can profoundly affect the ecology and conservation of species (e.g. Pulliam & Danielson 1991; Donovan et al. 1995a; Kawecki 2004). To date, the effects of habitat on population structure (e.g. Fryxell 2001; Reid et al. 2006) and species viability have been extensively modelled (e.g. Akçakaya et al. 2004; Larson et al. 2004). Some theoretical studies have also examined the coexistence of species and competitor, predator or parasite enemies (e.g. May & Robinson 1985; Namba, Umemoto & Minami 1999; Barabás et al. 2004), or estimated their effect on wild populations by using habitat fragmentation as an index of enemy impact (e.g., Temple & Cary 1988; Donovan et al. 1995b; With & King 2001; Lloyd et al. 2005). It remains to be shown, however, whether empirical models of host and enemy abundance, land cover and demography could predict host population trends in nature.

We present empirical data and spatial models to describe the demography of an archetypical passerine, the song sparrow Melospiza melodia, in relation to a key enemy, the brood parasitic brown-headed cowbird Molothrus ater. Specifically, we estimated expected population growth rate for sparrows in the Southern Gulf Islands, BC, Canada, using data from 17 populations studied for 1–36 years from 1960 to 2006 and subject to varying levels of parasitism. Much evidence shows that cowbirds can limit population growth in song sparrows via their effect on host reproduction (reviewed in Arcese & Marr 2006). Cowbirds are also implicated in the decline of other hosts via brood parasitism (Mayfield 1977; Robinson et al. 1995; Kus & Whitfield 2005) and nest depredation (Arcese, Smith & Hatch 1996; Hoover & Robinson 2007), particularly after invading new regions as a consequence of land conversion (Rothstein 1994). Smith et al. (1996) suggested that song sparrow populations subject to 60–80% parasitism acted as population sinks, but became self-sustaining after cowbird abundance was reduced experimentally (Smith, Taitt & Zanette 2002). Because cowbird abundance is linked to the distribution of feeding areas (reviewed in Chace et al. 2005), we expected that source–sink dynamics might develop as a consequence of spatial variation in the intensity of parasitism (Wilson & Arcese 2006).

Two main questions drove our study. First, we asked if spatial limits on cowbird distribution have the potential to create host refuges. We next asked if we could estimate the regional effect of cowbirds by mapping the predicted growth rates of local host populations. Based on earlier results, we predicted that host populations farther from cowbird feeding areas act as refuges from parasitism and achieve higher reproductive rates than populations with abundant cowbirds. To address our questions, we used field surveys, geographical information systems (GIS), and species habitat and population models to (i) predict parasitism rates based on landscape features influencing cowbird distribution; (ii) predict song sparrow distribution; (iii) relate sparrow demography to cowbird parasitism; and (iv) map potential population growth rates for song sparrows based on cowbird distribution. Last, we attempted to validate our predictions using empirical census data from focal populations and the North American Breeding Bird Survey (BBS).

Materials and methods

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

study area

We studied cowbirds and song sparrows in a 27 × 57 km portion of the Southern Gulf Islands (SGI), BC, Canada (Supplementary Material Fig. S1), where vegetation falls mainly into the coastal Douglas fir Psuedotsuga menziesii biogeoclimatic zone, and where many smaller islands are vegetated by dense shrub, and elevation ranges from 0 to 600 m above sea level. The study area included > 100 islands of 0·01–186 km2 in size, and was classified as approximately 70% forested, 13% rural, 6% agriculture, 2% suburban, and 9% other (Islands Trust 2002).

study species

Song sparrow

Song sparrows in our study area are common in shrub, riparian, and woodland habitats, raise one to four broods annually, reside year-round on territories approximately 200 to 5000 m2 in size, and typically breed for 1 to 4 years (Arcese et al. 2002). When parasitized, song sparrows often rear cowbirds and their own young successfully but nest failure rises linearly with the presence and intensity of parasitism (Arcese & Marr 2006).

Brown-headed cowbird

Cowbirds parasitize hosts mainly in forest edge and savanna habitats but feed in pastures and lawns (Rothstein, Verner & Stevens 1984; Thompson 1994; Chace et al. 2003). Distance to foraging area can restrict cowbird distribution (e.g. Tewksbury et al. 1999; Goguen & Mathews 2001) and cause marked variation in parasitism in populations < 5 km apart (Wilson & Arcese 2006). Cowbirds reduce host reproduction by removing eggs, usurping parental care, and causing nest abandonment and depredation (e.g. Arcese, Smith & Hatch 1996; Hoover & Robinson 2007). On Mandarte Island, cowbirds reduce reproduction in female song sparrows mainly by causing nest failure (Arcese & Marr 2006).

species distribution models


We developed species distribution models for cowbirds and song sparrows based on 477 point counts conducted from 06.00 h to 11.00 h on 35 islands between 6 May and 10 July 2005. Playbacks for song sparrows, using a locally recorded song, effectively lengthened 10-min point counts to 13 min and increased detection probability. We used all detections ≤ 100 m from an observer, stratified count locations over all landcover classes in the GIS, spaced counts by ≥ 200 m, and recorded count location (GPS60, Garmin Ltd, Olathe, KS, USA; see also Jewell, Arcese & Gergel 2007).

Habitat models

We adopted the species distribution model for brown-headed cowbirds of Jewell, Arcese & Gergel (2007), who detailed its development and validation. Briefly, we reduced an a priori list of candidate predictors and models of cowbird occurrence from GIS layers representing vegetation cover and distance to, and proportion of, feeding areas and forested areas in the landscape (Supplementary Material Table S1). We used forward stepwise logistic regression to find a ‘best’ cowbird model. We corrected for spatial autocorrelation in occurrence and environmental data using autologistic regression and a modified Gibbs sampler (following Augustin, Mugglestone & Buckland 1996); this method adds an extra term, the autocovariate, to the logistic model, which modifies the predicted probability of occurrence in a cell by values in neighbouring cells. It is well known that autologistic regression reduces bias in species distribution models due to spatial autocorrelation, and increases predictive power (e.g. Osborne, Alonso & Bryant 2001; Lichstein et al. 2002; Betts et al. 2006). We chose a spatial neighbourhood with a 1200-m radius as the appropriate scale for the autocovariate by comparing Akaike Information Criterion scores for models with different neighbourhood sizes (Burnham & Anderson 2002).

We used Nagelkerke's R2, which approximates the R2 of least squares (Nagelkerke 1991), to estimate the amount of variance explained. Model discrimination was assessed using receiver-operating characteristics plots, where the area under the curve (AUC) indicates overall model fit, 0·5 suggesting even odds and 1·0 implying perfect discrimination.

We followed Jewell, Arcese & Gergel (2007) to model song sparrow distribution based on an a priori set of candidate variables available in a GIS (Table S1). Predictive distribution maps for sparrows and cowbirds were created by projecting autologistic models as a combination of input layers in a raster GIS (25 m resolution) to estimate probability of occurrence in each pixel. Because the habitats we surveyed were generally open and both species were highly active and vocal during sampling, we did not adjust probability of occurrence for probability of detection. We do, however, validate our predictions using empirical data in Jewell, Arcese & Gergel (2007).

We linked our song sparrow distribution map to our population model by assuming that sparrows were present in all cells with a probability of occurrence ≥ 0·70. This cut-off matched the prevalence of song sparrows in point counts and satisfied the equality of sensitivity (rate of true positives) and specificity (rate of true negatives) of predictions (Liu et al. 2005).

creating a parasitism map

Field data

We estimated parasitism rates using our cowbird distribution map and empirical data from 12 island and three mainland sparrow populations studied for 1–8 years from 1995–2005 (Table 1; for details, see Smith et al. 2006; Wilson & Arcese 2006). Briefly, sparrows were individually marked and their territories visited weekly from April to July to monitor the fate of all nests. For each nest, date of first egg (DFE), number of cowbird eggs, and number of sparrow young fledged were noted. Annual parasitism rate at each site equalled the fraction of nests with ≥ 1 cowbird egg. To estimate an observed probability of occurrence for cowbirds, we conducted repeat point counts (mean = 16) in all 15 study sites in 2005–2006.

Table 1.  Data used to relate observed probability of cowbird occurrence to parasitism rates experienced by song sparrows. Parasitism rates were derived from 15 populations, controlling statistically for number of nests and inter-annual variation
SiteN (years)Total nestsParasitism*Mean occurenceSize (km2)
Mean95% CI
  • *

    means with asymmetrical confidence limits (original units; per Sokal & Rohlf 1985).

  • mean cowbird occurrence estimated over multiple site visits (sample size).

  • subset of sites used to estimate reproductive output for song sparrows (‘demographic data’). Rubly (N = 6 years) and Strawberry (N = 2 years) were also included.

  • §

    mainland sites.

Darcy2 410·110·20–0·830·27 (11)8·9
Dock 18 690·460·30–0·620·54 (13)0·1
Dock 28 630·410·26–0·570·85 (13)0·09
Dock 36 240·120·02–0·300·00 (6)0·05
Imrie5 27< 0·010·06–0·050·17 (12)0·1
Ker81070·470·32–0·610·41 (22)0·6
Little Shell7 710·430·27–0·600·31 (13)0·1
Mandarte84100·01< 0·01–0·040·00 (8)0·8
Reay8 470·120·03–0·250·25 (12)0·2
Sidney4 270·030·01–0·190·11 (27)117·5
Piers2 240·340·01–1·000·56 (25)12·0
Rum1 110·020·06 (16)0·6
Westham§32370·690·46–0·880·85 (20)2·0
Deas§21110·790·17–0·970·70 (20)2·5
Delta§31480·550·30–0·780·40 (20)1·8
Statistical analyses

We used analysis of variance (anova) to estimate the mean rate of parasitism at each site over all years, controlling statistically for breeding date (Wilson & Arcese 2006). Parasitism was normalized (arcsine) and the analysis was weighted by the square root of the total number of nests to reduce estimation error (cf. Arcese et al. 1992). We regressed the average parasitism rate recorded on the probability of cowbird occurrence (from repeat point counts) in all 15 study sites (variables normalized, regression forced through the origin). We used this regression to transform our cowbird distribution map into a map of the expected rate of parasitism in song sparrows across our study area.

estimating reproductive output

We estimated reproductive output based on the expected rate of parasitism in each map pixel, and then combined those estimates with empirical estimates of juvenile and adult survival to estimate population growth rate (λ). Modelling seasonal fecundity is correct for our purpose because it captures the influence of parasitism on total reproductive output (Arcese & Smith 1999; Grzybowski & Pease 2000). To estimate reproductive output under parasitism, we used 34 years of data from Mandarte Island to find the most parsimonious model of reproductive output (cf. Arcese & Marr 2006). We then fit this model to our multi-island data to estimate reproductive rate in 10 sites (Table 1), using ‘island’ as a random factor, an autoregressive covariance structure, and three fixed effects: parasitism, median DFE in each population annually, and the number of females breeding (PROC MIXED, sas 9·0, SAS Institute Inc., Cary, NC, USA). Because cowbirds usually begin breeding in May, whereas song sparrows begin in March or April, more nests escaped parasitism and reproductive output was higher on average as DFE declined (e.g. Smith et al. 2006). In the absence of parasitism, reproductive output increased linearly with the number of females breeding on Mandarte Island in 1960–1963 and 1975–2006, but the opposite relationship was seen in years with cowbirds present due mainly to cowbird-related nest failure (Arcese & Marr 2006). Our final model to predict the number of locally-produced fledglings (log10) used observed parasitism rate, DFE, and the number of breeding females (log10).

We used this final model to estimate reproductive output for song sparrows in each pixel of the study area while holding covariates constant. To do so, we assumed all populations shared a DFE equal to the mean of all study years (103·9, range: 79–126, n = 60; Julian day), and that each occupied pixel had a population size equal to two females per 25-m pixel (625 m2), based on the median size of territories on Mandarte Island over 31 years (357·3 m2, range: 35·4–5067·7, n = 1218 males).

estimating population growth rate

We estimated the expected growth rate of sparrow populations in the absence of immigration from the equation:

  • λa = Sf + (Nj * Sj)(eqn 1)

where Sf  is the apparent survival of breeding females, Nj the number of female fledglings produced per female assuming a 1:1 sex ratio at fledging, and Sj the apparent survival of fledglings from nest-leaving to breeding age (1 year). Nj was estimated from our model of reproductive output (above). Our use of λa emphasizes that our estimates of population growth rely on apparent survival, ignore dispersal and may therefore under- or overestimate true population growth rates in nature. We therefore compared these estimates to estimates of population growth based on long-term censuses, and we discuss potential biases in the Discussion. Values of λa ≥ 1·0 indicate stable to growing populations; λa < 1·0 indicates a declining population.

Apparent survival underestimates actual survival because it ignores the rate of permanent emigration by adults (Sandercock 2006). However, much work shows that permanent adult emigration is very rare in our system (e.g. Smith et al. 2006); thus, ‘apparent survival’ should be close to true rates in nature. We estimated apparent survival in adult female sparrows (Sf: 0·50 ± 0·32 SD, range: 0–1·00, n = 59island-years) from empirical observations on 10 islands studied for 2–8 years, assuming a detection rate of 1·0 based on a re-sighting rate of 0·99 (0·98–1·00 95% CI; Wilson & Arcese 2008). By contrast, empirical estimates of apparent juvenile survival were very low except on Mandarte Island (mean Sj: 0·05, range: 0–0·68, n = 59island-years). We therefore assumed a study-wide Sj = 0·25, following a common assumption for north-temperate passerines (Greenberg 1980). This is less than the mean for Mandarte Island (0·31 ± 0·19, n = 31 years) but higher than estimates from other sites (Arcese et al. 1992).

We validated our estimates of λa by comparing predicted values to independent estimates based on (i) trends from longitudinal censuses of populations in our study area, and (ii) regional trends from BBS routes within 100 km of the centre of the study area and with ≥ 10 years of data (Sauer, Hines & Fallon 2006). Trends were estimated on islands from censuses of all territorial birds at the end of April each year or regionally as the weighted average of trends (square-root, number of censuses). Finally, we evaluated the sensitivity of our estimates of λa in populations subject to varying rates of parasitism to variation in adult and juvenile apparent survival and the strength of parasitism's effect on reproduction (Supplementary Fig. S2).


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

modelling distribution and parasitism

Our habitat model predicted that cowbirds occurred closer to urban areas, preferred areas with more cattle grazing, avoided agriculture, and preferred some landcover types (i.e. rural, wetlands) over others (i.e. mature forest; see Supplementary Material Table S2). Our model described a modest amount of variance in cowbird occurrence (Nagelkerke R2 = 0·28) but discriminated well between occupied and unoccupied sites (AUC = 0·78). Our song sparrow model suggested that sparrows preferred to be closer to sea water but farther from fresh water, and preferred more open forests and areas with a high heterogeneity of landcover types (Supplementary Material Table S2). The best autologistic model for sparrows was similar to our cowbird model in variance explained (Nagelkerke R2 = 0·24) and discrimination (AUC = 0·75). Based on our 0·70 probability of occurrence cut-off, we predicted song sparrows to occur in 50% (191 km2) of our landscape.

Parasitism varied widely across our intensive study sites (mean: 0·32 ± 0·03, range: 0–0·81, n = 83island-years), with site and season length accounting for 79% of the total variance (anova,‘site’: F15,66 = 16·74, P < 0·001; ‘season length’, F1,66 = 10·45, P = 0·002). Expected mean parasitism rate at each site increased linearly with the odds of detecting cowbirds in point counts (r2 = 0·87, SEE = 0·23, F1,14 = 97·29, P < 0·001; Fig. 1). Thus, we used the slope of this regression (β = 0·828) to transform our cowbird distribution map into a predictive parasitism map (mean: 0·15 ± 0·12, range: 0–0·79; Fig. 2).


Figure 1. The mean parasitism rate of song sparrows increased linearly with observed probability of cowbird occurrence in 15 intensive study sites (original units).

Download figure to PowerPoint


Figure 2. Map of expected cowbird parasitism rates of song sparrows.

Download figure to PowerPoint

reproductive output

Parasitism contributed significantly to variation in reproductive output across sites (t = −2·37, P = 0·022; ‘DFE’: t = −4·38, P < 0·001; ‘females’: t = 10·00, P < 0·001). This mixed model differed from a null model (χ2 = 6·16, d.f. = 1, P = 0·013), and likelihood ratio tests indicated that including ‘island’ as random factor improved model fit (χ2 = 3·9, d.f. = 1, P < 0·05). Our final model suggested that reproductive output can be estimated as:

  • Y = 1·78 + 0·90 (log[females]) – 0·01 (DFE) – 0·31 (parasitism) (eqn 2)

Thus, we predict that song sparrows produce 4·60 young per female in the absence of parasitism on average, but only 2·42 young per female where parasitism is highest. Overall mean predicted reproductive output (4·02 ± 0·41) was very close to empirical values recorded on Mandarte Island (4·01 ± 0·27, n = 31 years) and all islands (3·95 ± 0·28, n = 60island-years).

predicted population growth rate

At the regional scale, the mean predicted λa suggests a stable population and the existence of population sinks where parasitism is expected to be high, but no strong sources (Fig. 3; mean λa: 1·00 ± 0·05 SD, range: 0·80–1·08). Locally, populations are expected to grow in the absence of immigration with cowbirds absent, but to decline where parasitism exceeds approximately 20%. Considering only our 27 km2 demographic study area (Supplementary Material Fig. S1), our map predicts population declines (7·8% per year) in the absence of immigration (mean λa: 0·92 ± 0·06 SD).


Figure 3. Map of expected population growth rates for song sparrows considering the impact of parasitism. ‘No song sparrows’ applies to where sparrow occurrence was < 0·70 (probability of occurrence cut-off). No strong sources (λa > 1·10) were predicted.

Download figure to PowerPoint

Estimates corresponded well to observed rates of λa in sparrow populations. Observed rates of decline on six islands studied intensively for 8 to 31 years averaged 2·3% (± 3·6 SE), which is within 1 SD of the predicted λa for the demographic study area. Eighty per cent of 15 BBS routes within 100 km of our study area also showed negative trends (mean: −2·5%± 0·7 SE; Supplementary Material Table S3).

Estimated population growth rate was sensitive to assumptions about adult and juvenile apparent survival (Supplementary Material Fig. S2). At the mean rate of parasitism predicted for song sparrow populations (0·18), varying either adult or juvenile apparent survival ± 1 SD changed our estimate of λa by 32%. Variation in juvenile apparent survival had its greatest effect when parasitism was low. Because apparent survival may underestimate true survival due to undetected dispersal, our sensitivity analysis also implies that predicted population growth may be underestimated when true survival exceeds apparent survival. Varying parasitism ±1 SE from the mean rate caused λa estimates to vary by ±4%, but, at the maximum parasitism rate (0·80), varying the strength of the effect of parasitism (β) by ±1 SE changed estimates of λa by 10%.


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

patterns in population growth rate

Evidence suggests that source–sink dynamics occur in many species (i.e. Harrison & Taylor 1997), often as a consequence of predation or parasitism (Namba, Umemoto & Minami 1999; Barabás et al. 2004), but few studies have shown empirically how landscape structure and enemy distribution affect the spatial demography of populations in nature. Our results strongly suggest that the distribution and intensity of parasitism by brown-headed cowbirds affect host population trend locally and regionally via its effect on reproduction, and that the intensity of parasitism can determine if populations act as sources (λa > 1) or sinks (λa < 1). Our models suggest that prior to about 1960 when cowbirds colonized our study area, song sparrow populations more often grew to limits imposed locally by territoriality, contributed more emigrants on average, and experienced higher rates of re-colonization following extirpation (Arcese & Marr 2006).

Because spatial variation in parasitism is driven by the distribution of cowbird feeding areas, especially urban areas and pastures (Goguen & Mathews 2000; Chace et al. 2003; this study), our results also suggest that land use patterns will affect the spatial dynamics of other popular cowbird hosts. In song sparrows, populations far from cowbird feeding areas had the highest expected growth rates; whereas lower λas were expected in a central band through our study area where hosts resided adjacent to livestock and urban areas (Fig. 3). Predicted source populations were concentrated in the north of our study area, isolated from landscape features favoured by cowbirds. Overall, our predictions are consistent with experiments showing that host reproductive output rises when cowbirds and parasitism rates are reduced (Smith et al. 2002; Kus & Whitfield 2005; Hoover & Robinson 2007). These results further suggest that the persistence of vulnerable host populations will be increased in landscapes that limit the distribution of cowbird feeding areas to maintain host refuges from parasitism.

At the scale of the SGI, our results suggest a stable regional sparrow population wherein weak sources over much of the area supplement sinks in habitats available to cowbirds (Fig. 3). In our demographic study area, our predicted 8% annual decline exceeded the 2·3% decline estimated from censuses of enumerated island populations. In contrast, BBS surveys suggest long-term declines up to −6·3% per year, particularly in areas with high rural-residential cover (Supplementary Material Table S3). Overall, these results match closely trends predicted by our spatial model in the absence of dispersal. Moreover, because Wilson & Arcese (2008) showed that juvenile dispersal is common and driven by density-dependent processes in our study area, differences in our predicted and observed rates of population growth can be reasonably explained by assuming that populations farther from cowbird feeding areas act as host refuges from parasitism and thus ameliorate declines in population sinks via dispersal.

modelling source–sink dynamics

Few studies have shown empirically how the spatial demography of populations is affected by the distribution of enemies or other threats related to land use. Donovan et al. (1995b) approached this question by adopting habitat fragmentation as an index of enemy impacts on forest-dwelling songbirds and predicted that, regionally, populations will decline as sources become fragmented. Lloyd et al. (2005) employed a similar approach to map population growth in two songbird species in relation to forest cover. Other studies have combined habitat models and demographic data to estimate human impacts on wolf Canis lupus populations (Carroll et al. 2003; Hebblewhite & Merrill 2008) and predation by mink Mustela vison on water voles Arvicola terrestris (Rushton et al. 2000). Our study advances these approaches by demonstrating how robust empirical models of spatial variation in brood parasitism can be used to make spatial predictions of local and regional population trend and inform managers about land use patterns likely to enhance species conservation.

A common finding of studies like ours is that the landscape-level ratio of source-to-sink populations critically affects predicted population trend, and that refuges help to ameliorate declines overall via dispersal. Our results also support Brawn & Robinson's (1996) suggestion that trend estimates that ignore the potential for immigration from undocumented sources can hide real risks to regional population viability. Factors potentially tipping the balance towards regional decline and also potentially under the control of managers include increases in the ratio of sink-to-source populations (Vuilleumier & Possingham 2006) or habitat fragmentation as it affects enemy distribution and the presence of ‘ecological traps’ (Donovan et al. 1995b; With & King 2001).

limitations of our approach

We followed Donovan et al. (1995a,b), Lloyd et al. (2005), and others to create static, predictive maps of population growth in the absence of dispersal. In nature, however, immigration by juveniles is common in declining and less-isolated song sparrow populations (Wilson & Arcese 2008). Moreover, empirical (Wilson & Arcese 2006, 2008) and theoretical (Arcese & Marr 2006) studies suggest that immigration accelerates population growth and reduces extinction risk, as predicted theoretically for patchy populations linked by dispersal (e.g. Brown & Kodric-Brown 1977). These results suggest that realized growth rates in predicted sink populations in our study area may often be increased by emigrants from potential sources.

Our results are also sensitive to assumptions about adult and juvenile survival (Supplementary Material Fig. S2). In particular, our use of apparent survival has the potential to underestimate true survival, and thus, population growth, to the degree that permanent emigration is ignored. In the case of adult song sparrows, however, a very high re-sighting rate (0·99; Wilson & Arcese 2008) and long monitoring history (Smith et al. 2006) suggest that this potential bias is small because adults are resident year-round and extremely sedentary overall. Using apparent survival to model spatial dynamics in more vagile species, however, will confound attempts to quantify the effects of survival and emigration or to categorize populations as sources or sinks, and thus reduce the reliability of predictions about metapopulation persistence and conservation plans (Runge, Runge & Nichols 2006). Unfortunately, robust empirical estimates of survival are rare due to problems related to dispersal, migration, and re-sighting (Sandercock 2006). As a result, our empirical estimates of juvenile survival varied widely, and we therefore followed others who estimated Sj as 50% of Sf (e.g. Greenberg 1980; Temple & Cary 1988; Donovan et al. 1995a,b; Lloyd et al. 2005).

Other potentially influential factors included as assumptions here include: an equal sex ratio at fledging; a bounded study area (immigrants from outside the study area may also influence demography); no effect of habitat quality, predators other than cowbirds, inbreeding, or age on reproductive output (but see Arcese et al. 1992; Arcese & Smith 1999; Smith et al. 2006); detection probabilities in point counts equal to 1·0; and species distributed similarly from year-to-year. The last two assumptions may cause us to misrepresent the prevalence of sparrows and cowbirds in our landscape over the long-term. In contrast, our close correlation of parasitism rate and cowbird prevalence (Fig. 1) suggests that our methods were sufficient to achieve our main goal of estimating how land use and parasitism affect population trend in song sparrows. An additional caveat is warranted regarding spatial autocorrelation. We employed a leading method to account for autocorrelation in spatial data, but newer alternatives also exist (e.g. Hoeting et al. 2006). Jewell, Arcese & Gergel (2007) discussed the validation of autologistic models in detail and reported that our approach was statistically robust and computationally efficient using widely available mapping tools. We suggest that additional approaches be explored where autocorrelation presents more severe problems.

Finally, we stress that our current predictions provide long-run estimates of population trend to guide land management and scenario analysis. Using our maps to make fine-scale predictions of local population growth is not warranted without additional validation to account for dispersal, habitat associations, and ground truthing. Nevertheless, we agree with Anders & Marshall (2005) that reliable estimates of population trend based on detailed empirical study can provide essential guidance in conservation planning. Validating our model predictions against empirical observations of population trend at local and regional scales increased our confidence that our predictions are reliable.

conservation implications

Our results suggest that brood parasitism by brown-headed cowbirds drives spatial variation in population performance in the song sparrow, a common host, and that these patterns differ from those existing prior to cowbird invasion (ca. 1960). Our results further suggest that management actions to reduce cowbird impacts in proximity to potential host refuges is likely to enhance host persistence overall, and that maps like ours could support such planning efforts. Thus, our results should help guide conservation in landscapes where managers can influence parasitism rate via strategic land use planning, landowner incentives, and habitat acquisition or restoration. In our study area, the marked effect of urban areas and livestock on parasitism suggests that limiting the proximity of these land uses to source populations will conserve their value as host refuges. Because parasitism reduces reproductive output in many hosts (e.g. Robinson et al. 1995; Hoover & Robinson 2007) and the well-studied song sparrow has a similar life history as a number of other hosts, our models also offer a reasonable starting place for prospective analyses of host decline in species with poor empirical data. Further, by advancing the approaches previously used by others (e.g. Donovan et al. 1995a,b; Rushton et al. 2000; Lloyd et al. 2005), our approach provides a promising framework for exploring regional-scale spatial dynamics of species in relation to their key enemies or other spatially distributed threats in other contexts.


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

We thank L. Sampson, M. Janssen and M. Sloan for help in the field; S. and A. Wilson, D. Irwin, and two anonymous reviewers for comments; Islands Trust, Capital Regional District, and M. Wulder for spatial data; and the Tsawout and Tseycum Bands, Parks Canada, and A. and H. Brumbaum for island access. We received generous funding from W. and H. Hesse, the Anne Vallée Ecological Fund, NSERC Canada, and Paetzold and Hoffmeister Fellowships from UBC and the Faculty of Forestry.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Akçakaya, H.R., Radeloff, V.C., Mlandenoff, D.J. & He, H.S. (2004) Integrating landscape and metapopulation modeling approaches: viability of the sharp-tailed grouse in a dynamic landscape. Conservation Biology, 18 (2), 52637.
  • Anders, A.D. & Marshall, M.R. (2005) Increasing the accuracy of productivity and survival estimates in assessing landbird population status. Conservation Biology, 19 (1), 6674.
  • Arcese, P. & Marr, A.B. (2006) Population viability in the presence and absence of cowbirds, catastrophic mortality, and immigration. In: Conservation and Biology of Small Populations: The Song Sparrows of Mandarte Island (eds J.N.M.Smith, L.F.Keller, A.B.Marr & P.Arcese), pp. 17592. Oxford University Press, New York.
  • Arcese, P. & Smith, J.N.M. (1999) Impacts of nest depredation and brood parasitism on the productivity of North American passerines. In: Proceedings of the 22nd International Ornithological Congress, Durban, South Africa (eds N.J.Adams & R.H.Slotow), pp. 295366. BirdLife South Africa, Johannesburg, South Africa.
  • Arcese, P., Smith, J.N.M. & Hatch, M.I. (1996) Nest predation by cowbirds and its consequences for passerine demography. Proceedings of the National Academy of Sciences of the United States of America, 93, 460811.
  • Arcese, P., Smith, J.N.M., Hochachka, W.M., Rogers, C.M. & Ludwig, D. (1992) Stability, regulation, and the determination of abundance in an insular song sparrow population. Ecology, 73 (3), 80522.
  • Arcese, P., Sogge, M.K., Marr, A.B. & Patten, M.B. (2002) Song sparrow (Melospiza melodia). In: The Birds of North America, No. 704 (eds A.Poole & F.Gill. The Birds of North America, Inc., Philadelphia, PA, USA.
  • Augustin, N.H., Mugglestone, M.A. & Buckland, S.T. (1996) An autologistic model for the spatial distribution of wildlife. Journal of Applied Ecology, 33, 33947.
  • Barabás, L., Gilicze, B., Takasu, F. & Moskát, C. (2004) Survival and anti-parasite defense in a host metapopulation under heavy brood parasitism: a source–sink dynamic model. Journal of Ethology, 22, 14351.
  • Betts, M.G., Diamond, A.W., Forbes, G.J., Villard, M.A. & Gunn, J.S. (2006) The importance of spatial autocorrelation, extent and resolution in predicting forest bird occurrence. Ecological Modelling, 191 (2), 197224.
  • Brawn, J.D. & Robinson, S.K. (1996) Source–sink population dynamics may complicate the interpretation of long-term census data. Ecology, 77 (1), 312.
  • Brown, J.H. & Kodric-Brown, A. (1977) Turnover rates in insular biogeography: effect of immigration on extinction. Ecology, 58 (2), 44549.
  • Burnham, K.P. & Anderson, D.R. (2002) Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach, 2nd edn. Springer, New York.
  • Carroll, C., Phillips, M.K., Schumaker, N.H. & Smith, D.W. (2003) Impacts of landscape change on wolf restoration success: planning a reintroduction program based on static and dynamic spatial models. Conservation Biology, 17 (2), 53648.
  • Chace, J.F., Farmer, C., Winfree, R., Curson, D.R., Jensen, W.E., Goguen, C.B. & Robinson, S.K. (2005) Cowbird ecology: a review of factors influencing distribution and abundance of cowbirds across spatial scales. Ornithological Monographs, 57, 4570.
  • Chace, J.F., Walsh, J.J., Cruz, A., Prather, J.W. & Swanson, H.M. (2003) Spatial and temporal activity patterns of the brood parasitic brown-headed cowbird at an urban/wildland interface. Landscape and Urban Planning, 64, 17990.
  • Donovan, T.M., Lamberson, R.H., Kimber, A., Thompson, F.R., III & Faaborg, J. (1995b) Modeling the effects of habitat fragmentation on source and sink demography of Neotropical migrant birds. Conservation Biology, 9 (6), 1396407.
  • Donovan, T.M., Thompson, F.R., III, Faaborg, J. & Probst, J.R. (1995a) Reproductive success of migratory birds in habitat sources and sinks. Conservation Biology, 9 (6), 138095.
  • Fryxell, J.M. (2001) Habitat suitability and source–sink dynamics of beavers. Journal of Animal Ecology, 70 (2), 31016.
  • Goguen, C.B. & Mathews, N.E. (2000) Local gradients of cowbird abundance and parasitism relative to livestock grazing in a western landscape. Conservation Biology, 14 (6), 186269.
  • Goguen, C.B. & Mathews, N.E. (2001) Brown-headed cowbird behavior and movements in relation to livestock grazing. Ecological Applications, 11 (5), 153344.
  • Greenberg, R. (1980) Demographic aspects of long distance migration. In Migrant Birds in the Neotropics: Ecology, Behaviour, Distribution, and Conservation (eds A.Keast & E.S.Morton), pp. 493504. Smithsonian Institution Press, Washington, D.C.
  • Grzybowski, J.A. & Pease, C.M. (2000) Relative influence of predation and brood parasitism on seasonal fecundity in passerine birds. In Ecology and Management of Cowbirds and Their Hosts (eds J.N.M.Smith, T.L.Cook, S.K.Robinson, S.I.Rothstein & S.G.Sealy), pp. 14555. University of Texas Press, Austin, TX, USA.
  • Harrison, S. & Taylor, A.D. (1997) Empirical evidence for metapopulation dynamics. In Metapopulation Biology: Ecology, Genetics, and Evolution (eds I.A.Hanski & M.E.Gilpin), pp. 2742. Academic Press, San Diego, CA, USA.
  • Hebblewhite, M. & Merrill, E. (2008) Modelling wildlife-human relationship for social species with mixed-effects resource selection models. Journal of Applied Ecology, 45, 834844.
  • Hoeting, J.A., Davis, R.A., Merton, A.A. & Thompson, S.E. (2006) Model selection for geostatistical models. Ecological Applications 16 (1), 8798.
  • Hoover, J.P. & Robinson, S.K. (2007) Retaliatory mafia behavior by a parasitic cowbird favors host acceptance of parasitic eggs. Proceedings of the National Academy of Sciences of the United States of America, 104 (11), 447983.
  • Islands Trust. (2002) Islands Trust Ecosystem Mapping. Islands Trust, Victoria, BC, Canada. Available at, accessed April 2008.
  • Jewell, K.J., Arcese, P. & Gergel, S.E. (2007) Robust prediction of species distribution: spatial habitat models for a brood parasite. Biological Conservation 140 (3–4): 25972.
  • Kawecki, T.J. (2004) Ecological and evolutionary consequences of source–sink population dynamics. In: Ecology, Genetics and Evolution of Metapopulations (eds I.Hanski & O.E.Gaggiotti), pp. 2344. Elsevier, Burlington, MA, USA.
  • Kus, B.E. & Whitfield, M.J. (2005) Parasitism, productivity, and population growth: response of least Bell's vireos and southwestern willow flycatchers to cowbird control. Ornithological Monographs, 57, 1627.
  • Larson, M.A., Thompson, F.R., Millspaugh, J.J., Dijak, W.D. & Shifley, S.R. (2004) Linking population viability, habitat suitability, and landscape simulation models for conservation planning. Ecological Modelling, 180 (1), 10318.
  • Lichstein, J.W., Simons, T.R., Shriner, S.A. & Franzreb, K.E. (2002) Spatial autocorrelation and autoregressive models in ecology. Ecological Monographs, 72 (3), 44563.
  • Liu, C.R., Berry, P.M., Dawson, T.P. & Pearson, R.G. (2005) Selecting thresholds of occurrence in the prediction of species distributions. Ecography, 28 (3), 38593.
  • Lloyd, P., Martin, T.E., Redmond, R.L., Langner, U. & Hart, M.M. (2005) Linking demographic effects of habitat fragmentation across landscapes to continental source–sink dynamics. Ecological Applications, 15 (5), 150414.
  • May, R.M. & Robinson, S.K. (1985) Population-dynamics of avian brood parasitism. American Naturalist, 126 (4), 47594.
  • Mayfield, H. (1977) Brown-headed cowbird: agent of extermination? American Birds, 31, 10713.
  • Nagelkerke, N.J.D. (1991) A note on a general definition of the coefficient of determination. Biometrika, 78, 69192.
  • Namba, T., Umemoto, A. & Minami, E. (1999) The effects of habitat fragmentation on persistence of source–sink metapopulations in systems with predators and prey or apparent competitors. Theoretical Population Biology, 56 (1), 12337.
  • Osborne, P.E., Alonso, J.C. & Bryant, R.G. (2001) Modelling landscape-scale habitat use using GIS and remote sensing: a case study with great bustards. Journal of Applied Ecology, 38, 45871.
  • Pulliam, H.R. & Danielson, B.J. (1991) Sources, sinks, and habitat selection: a landscape perspective on population dynamics. American Naturalist, 137, S50S66.
  • Pulliam, H.R. (1988) Sources, sinks, and population regulation. American Naturalist, 132 (5), 65261.
  • Reid, J.M., Bignal, E.M., Bignal, S., McCracken, D.I. & Monaghan, P. (2006) Spatial variation in demography and population growth rate: the importance of natal location. Journal of Animal Ecology, 75, 120111.
  • Robinson, S.K., Rothstein, S.I., Brittingham, M.C., Petit, L.J. & Grzybowski, J.A. (1995) Ecology and behavior of cowbirds and their impact on host populations. In Ecology and Management of Neotropical Migratory Landbirds (eds T.E.Martin & D.M.Finch), pp. 42860. Oxford University Press, New York.
  • Rothstein, S.I. (1994) The cowbird's invasion of the Far West: history, causes, and consequences experienced by host species. Studies in Avian Biology, 15, 30115.
  • Rothstein, S.I., Verner, J. & Stevens, E. (1984) Radio-tracking confirms a unique diurnal pattern of spatial occurrence in the parasitic brown-headed cowbird. Ecology, 65 (1), 7788.
  • 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. American Naturalist, 167 (6), 92538.
  • Rushton, S.P., Barreto, G.W., Cormack, R.M., Macdonald, D.W. & Fuller, R. (2000) Modelling the effects of mink and habitat fragmentation on the water vole. Journal of Applied Ecology, 37, 475490.
  • Sandercock, B.K. (2006) Estimation of demographic parameters from live encounter data: a summary review. Journal of Wildlife Management, 70, 150420.
  • Sauer, J.R., Hines, J.E. & Fallon, J. (2006) The North American Breeding Bird Survey, Results and Analysis 1966–2005. Version 6·2·2006. USGS Patuxent Wildlife Research Center, Laurel, MD, USA.
  • Smith, J.N.M., Keller, L.F., Marr, A.B. & Arcese, P., eds. (2006) Conservation and Biology of Small Populations: The Song Sparrows of Mandarte Island. Oxford University Press, New York.
  • Smith, J.N.M., Taitt, M.J. & Zanette, L. (2002) Removing brown-headed cowbirds increases seasonal fecundity and population growth in song sparrows. Ecology, 83 (11), 303747.
  • Smith, J.N.M., Taitt, M.J., Rogers, C.M., Arcese, P., Keller, L.F., Cassidy, A. & Hochachka, W.M. (1996) A metapopulation approach to the population biology of the song sparrow Melospiza melodia. Ibis, 138 (1), 12028.
  • Sokal, R.R. & Rohlf, F.J. (1985) Biometry, 3rd edn. W. H. Freeman, New York.
  • Temple, S.A. & Cary, J.R. (1988) Modeling dynamics of habitat-interior bird populations in fragmented landscapes. Conservation Biology, 2 (4), 34047.
  • Tewksbury, J.J., Martin, T.E., Hejl, S.J., Redman, T.S. & Wheeler, F.J. (1999) Cowbirds in a western valley: effects of landscape structure, vegetation, and host density. Studies in Avian Biology, 18, 2333.
  • Thompson, F.R. (1994) Temporal and spatial patterns of breeding brown-headed cowbirds in the Midwestern United States. Auk, 111 (4), 97990.
  • Vuilleumier, S. & Possingham, H.P. (2006) Does colonization symmetry mater in metapopulations? Proceedings of the Royal Society, Series B, 273, 163742.
  • Wilson, A.G. & Arcese, P. (2008) Sex-specific immigration and population structure in an avian island metapopulation. Journal of Avian Biology, 39, 344347.
  • Wilson, S. & Arcese, P. (2006) Nest depredation, brood parasitism and reproductive variation in island populations of song sparrows (Melospiza melodia). Auk, 123, 78494.
  • With, K.A. & King, A.W. (2001) Analysis of landscape sources and sinks: the effect of spatial pattern on avian demography. Biological Conservation, 100 (1), 7588.

Supporting Information

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

Fig. S1. Map of the study site.

Fig. S2. Sensitivity analysis for sparrow population growth rate.

Table S1. Predictive variables used in analyses of occurrence

Table S2. Final habitat model variable parameters

Table S3. Sparrow population trend estimates from BBS routes

Please note: Blackwell Publishing are not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

JPE_1503_sm_FigureS1.pdf54KSupporting info item
JPE_1503_sm_FigureS2.pdf94KSupporting info item
JPE_1503_sm_TableS1.pdf77KSupporting info item
JPE_1503_sm_TableS2.pdf35KSupporting info item
JPE_1503_sm_TableS3.pdf108KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.