Off-territory movement of male American Redstarts (Setophaga ruticilla) in a fragmented agricultural landscape is related to song rate, mating status and access to females
- First Online:
- Cite this article as:
- Churchill, J.L. & Hannon, S.J. J Ornithol (2010) 151: 33. doi:10.1007/s10336-009-0419-x
- 99 Views
Male songbirds often move off-territory to pursue extra-pair fertilizations. This movement represents a trade-off between paternity gain and loss and can be influenced by male quality and access to fertile females. Access to females could be reduced in fragmented landscapes that have small patches and low connectedness. We studied movement and extra-pair fertilization success of radio-tracked male American Redstarts (Setophaga ruticilla) in forest patches in an agricultural landscape in Alberta, Canada, over 2 years. Males spent an average of 18% of their time off-territory, mostly intruding onto adjacent territories and rarely moving between patches. They averaged 0.8 trips/h, with mean trip duration of 17 min and mean trip distance of 104 m. Less time was spent off-territory when their mate was nest-building and males intruded most often onto territories with nest-building females. Males with higher song rates and more nearby females intruded most onto other territories. Monogamous males in better condition with higher song rates spent the most time off-territory. However, males with more nearby females and higher local breeding synchrony spent the least time off-territory, suggesting these males face a trade-off between seeking extra-pair fertilizations and protecting against cuckoldry. Forest cover was not an important predictor of movement. Investment in off-territory movement did not predict extra-pair fertilization success or probability of cuckoldry. However, few tracked males achieved extra-pair fertilizations (1/22 tracked males vs 18/57 non-tracked males), possibly an artefact of low sample size or the effect of radio transmitters on female choice.
KeywordsAmerican RedstartFragmentationMovementSong rateExtra-pair paternity
Males of many temperate-forest-breeding songbirds move off their territories during the breeding season. Some movement allows males to expand territorial boundaries (Birkhead and Møller 1992) or acquire public information (e.g., information about conspecific habitat quality; Doligez et al. 2002). Most off-territory movement, however, is probably to pursue extra-pair copulations (Hanski 1992; Stutchbury 1998; Woolfenden et al. 2005) or polyterritorial polygyny (Secunda and Sherry 1991). Investment in off-territory movement likely represents a trade-off between gaining fertilizations and losing paternity to cuckoldry (Birkhead and Møller 1992). Therefore, the costs and benefits of movement should depend on attributes of individuals (Stutchbury 1998; Naguib et al. 2001), the seasonal timing of movement, and the availability of and access to potential extra-pair mates (Perreault et al. 1997; Norris and Stutchbury 2001; Fraser and Stutchbury 2004; Woolfenden et al. 2005).
Increased off-territory movement and intrusions onto other territories have been documented for old males (Kleven et al. 2005) and males with high song rates (Naguib et al. 2001); song being an honest signal of good past or current condition (Gottlander 1987; Nowicki et al. 2000; Foerster et al. 2002). Similarly, within-pair and extra-pair fertilizations are most often achieved by old males (Kempenaers et al. 1992; Perreault et al. 1997) in good body condition (Dyrcz et al. 2005). Polygynous males likely move off-territory less than monogamous males, since the former probably have increased time and energy demands related to guarding multiple mates (Secunda and Sherry 1991). When their mates are fertile (e.g., during the nest-building and laying stages, Birkhead and Møller 1992), males should invest more in mate guarding and territory defence (Westneat 1990) and less in off-territory movement. However, when breeding density and breeding synchrony are high, males face a trade-off between reduced off-territory movement to avoid cuckoldry (Birkhead and Møller 1992) and increased movement since many fertile potential extra-pair mates are available (Birkhead and Møller 1992; Stutchbury et al. 1997). In fragmented landscapes with small isolated patches (Fahrig 2003), availability of and access to potential extra-pair mates could be reduced (Morton 1992; Norris and Stutchbury 2001), increasing male movement to find additional females (Norris and Stutchbury 2001). Increased access to females could be especially important in fragmented landscapes with high levels of nest predation since extra-pair fertilizations could provide males with some reproductive success if their own nest is depredated (Perreault et al. 1997). On the other hand, male movement could decrease in fragmented landscapes if males are averse to crossing gaps in forest cover (Desrochers and Hannon 1997; Grubb and Doherty 1999; Norris and Stutchbury 2002). Finally, female behavior should influence the costs and benefits of male movement if females search off-territory for potential extra-pair mates (Neudorf et al. 2002; Norris and Stutchbury 2002) or choose mates based on male phenotypic attributes (e.g., plumage coloration; Kappes 2004; Stutchbury et al. 2005).
In this paper, we assess the reproductive costs and benefits associated with movement of male American Redstarts (S. ruticilla) breeding in forest patches in an agricultural landscape in central Alberta, Canada. Redstarts are sexually dichromatic Neotropical–Neartic migrants, and males exhibit delayed plumage maturation (male plumage superficially resembles that of females until a male’s second breeding season; Sherry and Holmes 1997). Redstart males engage in a mixed reproductive strategy, including social monogamy with high frequencies of extra-pair broods (59%: Perreault et al. 1997; 64%: Kappes 2004) and polyterritorial polygyny (5–16%: Secunda and Sherry 1991). Male age and song appear to be related to male quality or female choice (Lemon et al. 1992; Perreault et al. 1997). After-second-year (ASY) males with novel song types have higher pairing success, clutch sizes and fledging success than second-year (SY) males (Lemon et al. 1992), and older males are most frequently sires of extra-pair and within-pair young (Perreault et al. 1997).
The objectives of this study were to: (1) describe patterns of off-territory movement by males; (2) determine the influence of breeding stage, male attributes (age, song, body condition and mating status) and access to females (breeding density, breeding synchrony and forest cover) on movement investment; and (3) determine the relationship between movement investment and extra-pair fertilization success and probability of cuckoldry. We predicted that males should invest more in movement when they are at the lowest risk of cuckoldry (i.e., after the fertile period of their mate) of higher quality (i.e., older with higher song rates and in better body condition) monogamous and residing in areas with greater access to females (i.e., higher forest cover). Since high breeding density and breeding synchrony might simultaneously increase the opportunity for extra-pair fertilizations and the risk of cuckoldry, we made no prediction on the direction of these effects on movement investment. Finally, we predicted that males that invested most in movement would be most successful at siring within and extra-pair young, since either (1) they would encounter potential extra-pair females most frequently, or (2) because males that move the most would be higher quality males or preferred as mates by both pair- and potential extra-pair females (i.e., older males with higher song rates and in better body condition).
Study site and study population
During the study period, we monitored reproductive success of Redstarts in 19 breeding clusters (Fig. 1). From 2003 to 2004, age structure (juveniles and adults) and sex ratios were similar, but Redstart abundance decreased (breeding females: 44–30; territorial males: 53–28; S.H., unpublished data).
We captured birds in mist nets using playbacks of Redstart song and a mounted Redstart. We banded birds with colored and metal leg-bands and took a blood sample for DNA analysis from the brachial vein. Plumage attributes (males) and shape and color of retrices (females, Pyle 1997) were used to classify birds as SY or ASY. We weighed males to the nearest 0.1 g, and measured non-flattened wing chord to the nearest 0.1 cm. As part of a long-term demographic study (S.H., unpublished data) all territories in the study area were visited every 2 days to assess mating status (i.e., unpaired, monogamous or polygynous) and to locate and monitor all nests. Males were considered monogamous if a female nested on his territory (i.e., built a nest and laid at least one egg), polygynous if a male had two or more females nesting on his territory or held a second territory that contained a second nesting female, and unpaired if no female nested on his territory during the breeding season.
Eight and 14 different males were radio-tracked in 2003 and 2004, respectively (i.e., 22 individual males). We randomly chose males from within clusters that were contained in forest patches representing a range in local forest cover and breeding densities. Nineteen males were paired and two males in 2003 and one in 2004 were unpaired. A 0.43-g whip antenna radio transmitter (LB-2N; Holohil Systems) was attached with glue to the back [a technique similar to Raim (1978) but modified for use with Redstarts (C. Gillies, personal communication; S1 in Electronic Supplementary Material, ESM)]. Tracked males weighed between 6.8 and 9.2 g, so radio transmitters represented 5–6% of body mass.
Across years, tracking occurred between 1 June and 8 July (by week: 23, 52, 65, 27, 14 and 3 tracking sessions/week). Each male was tracked for 2 h once every 2–3 days [randomly assigned to a starting time of either 0600 (48 total sessions), 0800 (42 sessions), 1000 (49 sessions) or 1400 hours (30 sessions), MST], across the breeding stages [nest-building (60 sessions), laying (26 sessions), incubation (72 sessions) and nestling (11 sessions)] of his social mate, for the life of the transmitter battery (14–21 days; mean track time per male = 15.6 ± 4.2 h, range 8.6–24.5 h). The fewest tracking sessions occurred during the laying stage (since this is the shortest stage, i.e., 3–5 days) and the nestling stage (since radio transmitter batteries usually died before this). If off-territory movements extended longer than the 2 h session, males were tracked until they returned to their territories. We typically remained 5–20 m from males and, whenever they changed location or interacted with a predator or another bird, we recorded their locations, behavior (e.g., chases, displays) and whether they were singing, calling or silent. We mapped bird locations relative to 25-m-spaced grid points (within breeding clusters) or relative to flags placed during tracking sessions (outside breeding clusters). Grid points and flags were later surveyed with a handheld GPS. Provided we were within 30 m of the bird (most locations), we visually estimated that we were able to determine its position to within a 2.5-m radius of its true location (J.C., personal observations). Since differentially corrected GPS points had a mean horizontal precision of about 2 m, we estimate that the recorded location of each bird had a maximum error radius of 4.5 m (Churchill 2006).
Analysis of tracking locations and territory boundaries
ArcMap 9.0 (ESRI 2004) was used for all spatial analysis. Aerial photos from 2003 (1:20,000) were geo-referenced with 1:50,000 Alberta base features and digitized to create a landscape coverage classified into forest, agriculture, water, and anthropogenic features (i.e., roads and buildings). The minimum mapping unit was approximately 5 × 5 m. Tracking locations were plotted and converted to paths (Hawth’s Tools; Beyer 2004) and line lengths and distances from features were measured using the ArcMap measure function.
Territory boundaries were estimated so we could determine when males moved off-territory. For non-radio-tracked males, territory boundaries were mapped from locations of singing males taken about every 2 days from territory establishment until the nestling stage (>50 total locations per male). Territory maps were scanned, geo-referenced and digitized. We applied a 6-m buffer inside territorial boundaries to account for location errors in the field (i.e., estimated from error in identifying the bird’s true position plus error in transferring positions to 25-m grid scale maps). Buffers were placed inside territorial boundaries (as opposed to outside) to decrease our likelihood of falsely detecting intrusions onto a territory (see below). If the nest of a breeding pair fell outside the inner 6-m buffer, the unbuffered territory boundary was used for that pair (this happened in 19 of 74 cases). For tracked males, territory boundaries were derived from plotted singing locations (mean singing locations per male = 110 ± 59, n = 22 males). We used the fixed kernel density estimator in Hawth’s tools (h = 25, percent volume contour = 95) and removed points outside this. A minimum convex polygon was then created and a 4.5-m outer buffer was applied to account for the aforementioned error in radio-tracked bird locations. Minimum convex polygons were chosen to define territory boundary lines, rather than kernel density contours, since singing locations of males generally enclose defended areas.
Movement by a male off his territory was considered off-territory movement, outside a breeding cluster was off-cluster movement and outside a forest patch was off-patch movement. If the two territories of a polygynous male were adjacent, his movement between the territories was not considered to be off-territory. If a male moved onto the territory of another bird, this was called an intrusion, otherwise it was considered movement into unoccupied habitat. We calculated trip distance as net displacement from the territory boundary (i.e., the straight line distance).
Sample collection and DNA extraction
Complete genetic sampling during the study years was prevented by nest predation and parasitism by Brown-headed Cowbirds (Molothrus ater; because of egg ejection and killed chicks). A total of 333 samples were collected from 54 females, 66 males and 213 offspring. Sampling effort was equal in both years (2003: 63% of 176 eggs laid; 2004: 63% of 99 eggs laid). Between 25 and 50 μL of blood was collected from the brachial vein of adults and 6-day-old nestlings and stored in Queen’s lysis buffer (Seutin et al. 1991). DNA was also extracted from tissue of 38 chicks that died during rainstorms and 33 unhatched eggs. DNA extractions were conducted using QIAGEN DNeasy® Tissue Kits with specialized protocols for avian blood and tissue (Bush et al. 2005). DNA concentration ranged from 50 to 250 ng/μL.
DNA was amplified via polymerase chain reaction (PCR) with a 5′ fluorescent dye on either the forward or reverse primers in conditions optimized for each primer (Churchill 2006). We used six di-nucleotide-repeat-motif microsatellite primers, developed for use in other songbird species (Maμ 23: Alderson et al. 1999; Dpμ 03, Dpμ 16: Dawson et al. 1997; Cuμ 04: Gibbs et al. 1999; Dca 28, Dca 32: Webster et al. 2001). Microsatellite loci were highly polymorphic (range 5–18 alleles) showed high heterozygosity (range 0.500–0.940) and our microsatellite set gave high total exclusion probability (>0.998 in each year). We checked each locus for heterozygote deficiency each year using all adults in the study population that year (GenePop on the Web; Raymond and Rousset 1995). Only Dpμ 03 showed a significant heterozygote deficiency after a Bonferroni adjustment for multiple tests (Webster et al. 2004), but since it was highly polymorphic and deficient in only 1 year it was used in analyses.
Samples were run on polyacrylamide gels on an ABI Prism® 377 DNA sequencer, data were collected with GeneScan® analysis software and Genotyper® software (Applied Biosystems 2001a, b). A size standard was run in each lane to enable allele sizing. Samples that could not be scored the first time were re-run. Samples with allele sizes that did not conform to the expected di-nucleotide-repeat-motif (e.g., an allele of 163 bp when 162 and 164 bp were expected) were re-run to confirm their consistency.
We used Cervus 2.0 for paternity analysis which ranks putative fathers based on the statistical likelihood that they are the true father (Marshall et al. 1998). The likelihood approach accounts for allele frequency distributions, incomplete sampling of candidate sires, missing genotypes and mismatches caused by mis-typing, null-alleles and microsatellite mutation (Marshall et al. 1998). The number of unsampled males was estimated from the number of known unsampled males plus an allowance of two potential floaters or males residing beyond the study population boundaries (we rarely saw floaters but assume some were present that we did not sample). We assumed chicks were offspring of their social mother, and so the genotyping error rate (average = 0.02) was estimated from the number of mismatches between mothers and chicks (Marshall et al. 1998). We expected that if chicks were descended from their social mother they should show few allelic mismatches, but if they were the result of intra-specific brood parasitism we expected to see multiple mismatches (Webster et al. 2004). Only two offspring mismatched their social mother at two or more loci (both in 2003), and we assumed these to be potential incidences of intra-specific brood parasitism or mixed or mislabeled samples. When multiple nest-mates mismatched their mother at the same locus, we counted this as one mismatch and adjusted the genotyping error rate accordingly. We used the same error rate in simulations as in likelihood calculations and assigned paternity at 95 and 80% confidence levels.
We used a ‘total evidence’ approach to paternity assignment (similar to Webster et al. 2004). If the top ranked male mismatched the chick at two or more loci, we assumed that the true father was not among the pool of candidate males (since double errors within one genotype should be rare). A lower ranked male with fewer mismatches than the top ranked male was assigned as the true sire, if delta values were similar. When two males matched a chick at all loci, and the pair father was one of them, the pair father was assigned paternity. A sire was not assigned if two non-pair fathers had the same number of mismatches. We assigned a sire to more than 95% of analyzed offspring in all years. There was no significant difference in percentages of extra-pair young and extra-pair broods determined by Cervus, and through total evidence assignments or in the percentage of extra-pair offspring we assigned at 80 and 95% (S.H., unpublished data). Therefore, we conducted subsequent analyses using total evidence assignments and included young assigned at both confidence levels.
Correlates of movement investment
Our male attribute variables were age, mating status, body condition and song rate. Males were grouped by age category (ASY or SY), and by male minimum age [age; 1 (n = 8 males), 2 (n = 7) or 3+ which included one 4-year-old male (n = 7)] based on returns of banded birds. ASYs new to the study landscape were classified as 2 years old (Perreault et al. 1997). Mating status (status) was either unpaired (n = 3), monogamous (n = 16) or polygynous (n = 3; all were bigamous). Body mass increased with capture date [all males captured between 2002 and 2005 (S.H., unpublished data); capture date = 1 for date of first capture each year: r = 0.26, n = 125, P < 0.01], but was not related to time captured (0630–0829 hours, n = 29; 0830–1029 hours, n = 42; 1030–1230 hours, n = 27; 1330–1700 hours, n = 26: F3,120 = 1.20, P = 0.31). Therefore, we calculated a body condition index (condition) by regressing male mass on wing length (a surrogate of structural size) with capture date as a covariate. Increasingly positive and negative regression residuals represent males in better and poorer condition, respectively (Robb et al. 1992).
We calculated male song investment in two ways. First, we calculated the percent of on-territory locations where a tracked male was recorded singing (for sessions with more than five on-territory locations; song). Song rate was not related to the time of day a male was tracked; however, we excluded rainy days and sessions during the nestling stage when song rates were lower (Churchill 2006). Second, to enable comparisons among tracked and non-tracked males, in 2004, songs/min was calculated across three consecutive 1-min intervals each time a territory was entered for nest monitoring (except on rainy days and during the nestling stage).
We chose female breeding density, forest cover and breeding synchrony to represent the availability of and access to potential extra-pair partners for males. We predicted that male movement might be influenced by breeding females at the local scale (i.e., nearest neighbors) and at the neighborhood scale (within the breeding cluster or nearby breeding clusters). Thus, we measured breeding density as the number of breeding females in territories that at least partially fell within 200- and 500-m buffers of the tracked male’s territory centroid (density200 and density500) and forest cover as the percent forest within these buffers (forest200 and forest500). The fertile period of females was defined as the interval between 5 days before the first egg was laid and the day the penultimate egg was laid (Perreault et al. 1997). We calculated breeding synchrony based on the index of Kempenaers (1993) but used the average percentage of females fertile on a given day across the tracking period of a male within 200 and 500 m (synchrony200 and synchrony500).
In 4 clusters, no males were tracked; in 11 clusters, only one male was tracked; and in 4 clusters, multiple males were tracked. In two cases in 2004, two males from the same cluster were tracked, and in one case, three males from the same cluster were tracked. In a fourth cluster, one male was tracked in 2003 and three males were tracked in 2004. Clustered observations may not be independent, however, movement investment values were not spatially autocorrelated for tracked males (ArcMap; Moran’s I: all I < 0.17, all P > 0.10; Getis-Ord General G: all G < 0.01, all P > 0.10). Therefore, we treated each male as an independent observation. We also performed all statistical comparisons using mean values from each cluster, and results were not qualitatively different from those using all males individually, so we present only the latter analyses.
We used S-Plus 7.0 (Insightful Corporation 2005) for statistical analysis. For univariate analyses we used independent two-tailed t tests and report means ± SD. Dependent variables were highly correlated (all 0.50 ≤ r ≤ 0.89; S2 in ESM) so we used two dependent-variable selection methods. First, we retained intrusion frequency and time spent off-territory since they were the two least correlated variables (r = 0.50, P = 0.03) and because we felt that the costs and benefits to males of leaving the territory or intruding onto another territory should be different. Second, we retained maximum trip distance, which we thought best measured a male’s willingness or ability to move across the landscape. Since none of these measures differed by time of day males were tracked [Kruskal–Wallis rank test (n = 19); excludes unpaired males without permanent territories (see “Results”): intrusions/h: χ32 = 4.12, P = 0.25; time off-territory: χ32 = 7.10, P = 0.07; maximum trip distance: χ32 = 2.83, P = 0.42], data from all time categories were combined.
We ran multiple linear regression models for each dependent variable using scaled independent variables (by subtracting the mean and dividing by the SD; correlation matrix presented in S3 of ESM). For intrusion frequency and time off-territory, we specified global models with year, age, song, condition, status, density200, synchrony200 and forest200. For maximum trip distance, the global model was similar but used density500, synchrony500 and forest500 (instead of density200, synchrony200 and forest200) since long-distance movements might be influenced by resources at a larger scale (e.g., Norris and Stutchbury 2001; Woolfenden et al. 2005) and since most of male movements were within 500 m (see “Results”). We then screened global models for highly correlated terms to avoid collinearity. Age and song were highly correlated in all global models (0.6 < r < 0.8) so we compared global models to a model with song removed, and a model with age removed using AICc (Akaike Information Criteria with a correction for small sample sizes; Burnham and Anderson 2002), Akaike weights (wi, the weight of evidence, based on differences in AICc scores, that model i is the best model of the candidate set; Burnham and Anderson 2002) and evidence ratios (the likelihood of model i vs model j, based on the ratio of Akaike weights; Burnham and Anderson 2002). Models with age removed had much higher evidence ratios than models with song removed (range 2.6–49.0), so we present all models with age removed. We then performed supervised model reduction on these models using the step function in S-Plus to determine minimum adequate models (MAMs). Step is an iterative procedure that removes terms that when dropped result in a more parsimonious model (based on the Cp value which is parallel to AIC; Crawley 2002). If non-significant terms remained in the model following the Step procedure (i.e., P > 0.05), we removed them if their deletion did not significantly decrease explained variance of the model (ANOVA test, with α = 0.05; Crawley 2002).
Given that little is known about the strength and relative influence of male quality and female access variables on movement investment, we felt justified in using this more exploratory model selection approach. One monogamous ASY male was identified as an outlier in all models based on diagnostic plots (model residuals and Cook’s distance) and was therefore removed from all models (intrusions/h = 2.89, percent time off-territory = 52%, maximum trip distance = 1,170 m; S4 in ESM).
Comparison of tracked and non-tracked males and effects of radio transmitters
We first compared characteristics of tracked and non-tracked males to determine if tracked males were representative of other males in the population. Tracked males were similar to non-tracked males in age (proportion ASY: tracked = 0.64, n = 22, non-tracked = 0.61, n = 57; χ12 < 0.1, P = 0.85) and body condition (not available for all males: tracked mean = −0.17 ± 0.47, n = 21, non-tracked mean = −0.23 ± 0.64, n = 39; t58 = −0.4, P = 0.69). In 2004, tracked and non-tracked males had similar song rates (tracked: mean = 5.9 ± 2.1, n = 12, non-tracked: mean = 6.4 ± 2.7 songs/min, n = 5; t13 = 0.4, P = 0.70). We then compared return rates to the study area the subsequent year of tracked and non-tracked males to determine whether carrying a radio transmitter reduced male survival. Return rates following each study year were also similar (2004 returns: tracked = 25%, n = 8, non-tracked = 26%, n = 43; 2005 returns: tracked = 21%, n = 14, non-tracked = 14%, n = 14; years pooled: χ12 < 0.01, P = 0.99).
Only two males, both unpaired SY males, moved off-patch. Both males held territories for a few days and then roamed extensively across the landscape. Since this behavior differed from paired males, we did not include unpaired males in further analysis. Only 6 of 19 paired males moved off-cluster. Males that moved off-cluster tended to be closer to adjacent clusters than those that did not (mean = 328 ± 129 m, n = 6 vs mean = 552 ± 417 m, n = 13; t16 = 1.8, P = 0.10). All paired males moved off-territory at least once, mean trip distance was 104 ± 45 m (range 20–1,170 m) and 90 and 98% of trips were within 200 and 500 m, respectively.
Paired males (n = 19) moved off-territory an average of 0.8 ± 0.4 trips/h (range <0.1–1.8 trips/h), for an average of 17 ± 9 min/trip (range 1–110 min/trip), spending an average of 18 ± 1% of their time off-territory (range 0–52%; S4 in ESM). Paired males appeared to have two different strategies while they were off their territories. They either intruded silently onto the territories of other males (during 75% of 196 trips) or moved into unoccupied territory where they sang repeat songs (25% of 196 trips), which Redstarts use to attract mates (Sherry and Holmes 1997). In fact, 3 of 19 tracked paired males became polyterritorially polygynous: one male in the same cluster as his primary territory and two males in a different cluster in the same forest patch.
Timing of movements
Correlates of movement investment of paired males
Multiple linear regression models for three measures of male American Redstart (S. ruticilla) movement investment (n = 18 paired males)
The percentage of time males spent off-territory was significantly greater for monogamous males in good body condition, with high song rates and surrounded by females with low breeding synchrony. The exclusion of year in the MAM also revealed that males with a low local breeding density spent more time off-territory.
Finally, no terms in the global model significantly influenced the maximum distance travelled by males (Table 1). However, the MAM, with a maximal weight of evidence (wi = 1.00), revealed a significant effect of song (Table 1). Paired males with high song rates travelled the farthest from their territories. However, song rate only explained 35% of the variance in maximum distance travelled (Table 1).
Fertilization success in relation to movement and radio-tracking
We predicted that males investing more in movement would be more likely to gain extra-pair fertilizations. However, only 1 of 22 tracked males achieved an extra-pair fertilization (an unpaired SY male), so we could not test this. We also predicted that males that invested more in movement would be less likely to be cuckolded, but this was not the case. Forty percent of 15 tracked males with sampled nests were cuckolded but the likelihood of being cuckolded was not related to their frequency of intrusion onto other territories (not-cuckolded: mean = 0.81 ± 0.85 trips/h, n = 8, cuckolded: mean = 0.65 ± 0.56 trips/h, n = 5; t11 = 0.4, P = 0.70), percent time off-territory (not-cuckolded: mean = 16 ± 11%, n = 8, cuckolded: mean = 17 ± 10%, n = 5; t11 = −0.03, P = 0.97) or maximum trip distance (not-cuckolded: mean = 210 ± 153 m, n = 8, cuckolded: mean = 197 ± 123 m, n = 5; t11 = 0.17, P = 0.87).
Because so few tracked males achieved extra-pair fertilizations, we tested whether radio transmitters affected extra-pair fertilization success. Sample size and hence statistical power to detect a difference in fertilization success between tracked and non-tracked males was low. With years pooled, we found a lower proportion of tracked than non-tracked males achieved extra-pair fertilizations (2003: 13% of 8 tracked males, and 37% of 43 non-tracked males; 2004: 0% of 14 tracked males and 14% of 14 non-tracked; years pooled: tracked = 5%, n = 22, non-tracked = 32%, n = 57; χ12 = 5.0, P = 0.03). However, cuckoldry rates were similar for tracked and non-tracked males (2003: tracked = 67%, n = 6, non-tracked = 59%, n = 29; 2004: tracked = 22%, n = 9, non-tracked = 0%, n = 8; years pooled: tracked = 40%, n = 15, non-tracked = 46%, n = 37; χ12 < 0.1, P = 0.93).
Male Redstarts in our study area pursued a mixed reproductive strategy, similar to males in other populations (Secunda and Sherry 1991; Perreault et al. 1997). Off-territory movements were made either to intrude onto a female’s territory (75% of trips) or to sing in an unoccupied area, presumably to attract an additional female (25% of trips). During territorial intrusions, males were silent and were not observed foraging, suggesting that they were seeking extra-pair fertilizations or public information (Doligez et al. 2002) and not attempting to expand territory boundaries (Birkhead and Møller 1992). Below, we discuss timing of off-territory movements and the factors influencing these movements. In addition, we suggest three possible explanations for the puzzling result that only one tracked male achieved an extra-pair fertilization.
Timing of movement
Early in the season, when many females are nest-building simultaneously, males probably face a trade-off between guarding their mates and seeking extra-pair fertilizations (Westneat 1990). As predicted, males spent the least time off-territory during their mate’s nest-building stage, but contrary to our prediction, they spent the most time off-territory when their mates were laying and presumably still fertile (Birkhead and Møller 1992). This may be because female copulation frequency declines once the first egg has been laid (Ficken 1963; review in Birkhead and Møller 1992). Consistent with this, males also intruded most often onto territories of nest-building females. However, more than one-third of all intrusions were onto territories with incubating females. While these females were not fertile, males might have been searching for re-nesting females, since nest predation is high in this population (32% of all nests; S.H., unpublished data). Males with females that nested later than average were more likely to be cuckolded (J.C., in preparation), probably by males whose mates were already incubating. Males could also be intruding onto territories later in the breeding season to determine the reproductive success of neighbours, an indication of habitat quality in other areas of the breeding cluster (Doligez et al. 2002). Information on conspecific reproductive success might be important for Redstart males since they can switch territories within or among breeding seasons following nest failure (S.H., unpublished data).
Mating status and male quality
Male mating status should affect investment in off-territory movement if time and energy demands are higher for polygynous males. Although mating status was not a strong predictor of intrusion frequency or maximum trip distance, monogamous males spent more time off-territory than polygynous males (Table 1). This likely reflects higher time and energy demands for polygynous males to defend and move between their two territories. Consistent with this, in a 4-year study, only 8% of polygynous males achieved an extra-pair fertilization (relative to 14% of monogamous males; J.C., in preparation).
Of the male attributes we tested, song rate was consistently the strongest predictor of male movement investment: males with higher song intruded the most frequently onto territories of other males, spent the most time off their territories and moved furthest off-territory (Table 1). Male song rate is also positively correlated with off-territory movement investment in other species (Dark-eyed Juncos, Junco hyemalis: Chandler et al. 1994; nightingales, Luscinia megarynchos: Naguib et al. 2001; Reed Buntings, Emberiza schoeniculus: Kleven et al. 2005; but not Hooded Warblers, Wilsonia citrina: Stutchbury 1998). Song and movement are both energetically costly behaviors, so should be expressed most by males in good phenotypic condition (Sullivan 1994).
Although males with a higher body condition index (i.e., a higher mass relative to body size) spent more time off-territory than males with a low condition index, our body condition index was not consistently a strong predictor of male movement. Our index was not likely as good a measure of male condition as song rate since it did not directly account for a number of other factors contributing to condition: fat stores [which might most directly indicate body condition (Dyrcz et al. 2005), whereas mass can reflect both fat and non-fat body components (Rogers 2003)], changes in mass across the radio-tracking period (since males were only weighed once), differential wing feather wear among juvenile and adult birds [which could have caused inaccurate wing measurements and skewed relative condition among males, but see Francis and Wood (1989) and Merilä and Hemborg (2000)], immunocompetence (Saino et al. 1997) or parasite load (Dyrcz et al. 2005).
Availability of and access to females
Local breeding density was an important predictor of male intrusion frequency. Similarly, in a parallel 4-year study, we found that males on territories surrounded by more females were more likely to achieve extra-pair fertilizations (J.C., in preparation). High encounter rates could increase extra-pair fertilization success if females acquiesce to frequent harassment (Birkhead and Møller 1992). On the other hand, if females actively choose extra-pair mates, females who encounter more males could have an increased ability to compare them and select high-quality extra-pair mates (Sullivan 1994). Interestingly, males in areas of high breeding density and high breeding synchrony spent less time off-territory (Table 1). Together, these patterns suggest that males balance the costs and benefits of locating their territories in areas of high breeding density. When breeding density is high, males benefit by access to many fertile females, but might also risk cuckoldry if they leave their territories for long periods of time (Woolfenden et al. 2005; Stutchbury et al. 2005). Therefore, males in areas of high breeding density spend less time off-territory and presumably more time mate guarding or chasing away intruders (Westneat 1990) than males in clusters where breeding density is low.
Forest cover was not a significant term in any of our models, although it showed non-significant weak effect (i.e., small relative coefficient) in the most parsimonious model of time spent off-territory (Table 1). In our study area, forest cover likely has an indirect effect on functional connectivity since it is highly correlated with breeding density (S3 in ESM) and with breeding cluster dispersion [small breeding clusters have lower forest cover (cluster area vs percent forest cover within 200 m buffers of cluster centroids: t17 = −2.74, P = 0.01) and consequently fewer nearby clusters]. Therefore, our study design and low sample size may have precluded finding a stronger effect of forest cover on movement investment. An additional possibility is that male movement is not impeded by the level of forest cover in this landscape. In years of high breeding density, nearly half of males breeding in isolated fragments are cuckolded by floating unpaired males and ASY males from nearby large fragments (6 of 13 males across 2003 and 2004; J.C., in preparation). This suggests that some males are able to access all areas of the study landscape.
Extra-pair fertilization success of radio-tracked males
Tracked males had low extra-pair fertilization success and we propose three possible mechanisms to explain this: sampling error, female choice and radio transmitter effects. First, the fact that only 1 of 22 tracked males achieved an extra-pair fertilization over the 2 years could be an artefact of low sample size. Based on the percentage of non-tracked males that achieved extra-pair fertilizations in each year, we would only have expected three tracked males in 2003 and two tracked males in 2004 to achieve extra-pair fertilizations. Second, females might have rejected tracked males if they were of lower quality than non-tracked males. This was unlikely since we found no difference in age, body condition or song rates of tracked and non-tracked males. However, plumage characteristics are probably important in Redstart female choice (Kappes 2004) and we did not measure these. Third, negative effects of radio transmitters may have reduced extra-pair fertilization success. Males with radio transmitters may have invested less in movement and thereby encountered fewer females than males without radio transmitters, but we were unable to test this. Even if tracked males did not invest less in movement than non-tracked males, radio transmitters might have influenced female choice. Radio transmitters might influence female perception of male quality by altering his appearance or display ability. Adding structures such as colored leg-bands have influenced extra-pair fertilization success in some species (e.g., Zebra Finches, Taeniopygia guttata; Burley et al. 1994). Given the increase in studies of avian mating systems using radio-tracked males, it is essential to test the effect of radio transmitters on movement and female choice to appropriately interpret results.
Die Bewegung von männlichen Schnäpperwaldsängern (Setophaga ruticilla) außerhalb des Territoriums in einer fragmentierten Agrarlandschaft hängt mit Gesangsrate, Verpaarungsstatus und Zugang zu Weibchen zusammen
Männliche Singvögel verlassen oftmals ihr Territorium, um „extra-pair fertilisations” (Befruchtungen außerhalb des Paarbundes) nachzugehen. Diese Bewegung ist das Ergebnis eines Abwägens zwischen dem Gewinn und Verlust von Vaterschaft und kann durch die Qualität der Männchen und den Zugang zu fertilen Weibchen beeinflusst werden. Der Zugang zu Weibchen könnte in fragmentierten Landschaften mit kleinen Landschaftszellen und geringer Vernetzung reduziert sein. Wir haben die Bewegung und den Befruchtungserfolg außerhalb des Paarbundes von mit Radiosendern versehenen männlichen Schnäpperwaldsängern (Setophaga ruticilla) in Waldflecken in einer Agrarlandschaft in Alberta, Kanada, über einen Zeitraum von zwei Jahren untersucht. Die Männchen verbrachten im Durchschnitt 18% ihrer Zeit außerhalb ihres Territoriums, wobei sie zumeist in benachbarte Territorien eindrangen und sich selten zwischen den Waldflecken hin und her bewegten. Sie machten im Durchschnitt 0,8 Ausflüge pro Stunde mit einer mittleren Dauer von 17 min und einer durchschnittlichen Entfernung von 104 m. Die Männchen verbrachten weniger Zeit außerhalb des Territoriums, wenn ihre Partnerin mit dem Nestbau beschäftigt war, und Männchen drangen am häufigsten in Territorien mit nestbauenden Weibchen ein. Männchen mit höheren Gesangsraten und einer höheren Anzahl von Weibchen in der Nähe drangen am häufigsten in andere Territorien ein. Die längste Zeit außerhalb ihres Territoriums verbrachten monogame Männchen in besserer Kondition und mit höheren Gesangsraten. Männchen mit einer höheren Anzahl von Weibchen in der Nähe und höherer lokaler Brutsynchronität verbrachten indes die wenigste Zeit außerhalb ihres Territoriums, was darauf hindeutet, dass diese Männchen zwischen dem Suchen von außerpaarlichen Befruchtungen und der Absicherung gegen das „Fremdgehen” ihres eigenen Weibchens abwägen müssen. Die Walddecke war kein wichtiger Prädiktor für die Bewegung. Das Investment in die Bewegung außerhalb des Territoriums sagte nicht den außerpaarlichen Befruchtungserfolg oder die Wahrscheinlichkeit, selbst „betrogen” zu werden, vorher. Allerdings erreichten nur wenige besenderte Männchen außerpaarliche Befruchtungen (1 von 22 besenderten Männchen gegenüber 18 von 57 unbesenderten Männchen), was möglicherweise ein Artefakt der geringen Stichprobengröße oder der Effekt der Radiosender auf die Weibchenwahl sein könnte.
We thank the anonymous reviewers who provided comments on earlier versions of this manuscript. Thank you to C. McCallum, J. Manolo, A. Winkelaar, M. Hanneman, B. Harris, B. Laliberté and M. Sorensen for excellent field assistance and to C. Strobeck, W. Gallin, P. Dunn, M. Webster, L. Gibbs, D. Coltman, C. Davis, K. Bush and J. Bonneville for their genetics expertise and assistance. Thank you also to landowners who allowed us to work on their land and to the Meanook Biological Research Station staff for logistical support. This project was supported by the Natural Sciences and Engineering Research Council of Canada; the University of Alberta; the Alberta Conservation Association grants in Biodiversity; Alberta Sport, Recreation, Parks and Wildlife Foundation; Canadian Circumpolar Institute C/BAR grants; the Alberta Cooperative Conservation Research Unit; and a Walter H Johns graduate fellowship to J.C. This study complies with the current laws of Canada.