1Elevational gradients create environmental variation that is hypothesized to promote variation in life-history strategies. We tested whether differences in life-history strategies were associated with elevation in a songbird, the dark-eyed junco (Junco hyemalis; Aves; A.O.U. 1998).
2We monitored birds in four replicated sites per elevation, at 2000 m a.s.l. (high elevation) and 1000 m a.s.l. (low elevation), in the Rocky Mountains of Canada.
3Over 5 years, we measured the following traits and vital rates: egg-laying schedules, morphological indicators of reproductive stage, seasonal reproductive success, indicators of competitive class (age, size, arrival time), and survival rates.
4We found two main patterns: with an increase in breeding elevation, dark-eyed juncos delayed the development of structures necessary for reproduction (e.g. cloacal protuberance in males) and reduced the duration of their reproductive period to less than half of the time used by low-elevation birds; and
5Juncos at high-elevation sites had 55–61% lower annual reproductive success and 15 to 20% higher survival rates. While adult juncos at high elevations produced fewer offspring, those offspring were in better condition. Proportions of age and size classes in high- compared to low-elevation populations were similar, suggesting that a life-history trade-off is present, rather than competition forcing inferior competitors to breed in a peripheral habitat. The apparent trade-off between reproduction and survival corresponded to a shorter period of favourable weather and available food in high- compared to low-elevation habitats.
6Thus, elevation had a strong influence on life-history characteristics of a single species over a short spatial distance, suggesting a shift in life history from a high reproductive strategy at lower elevations to a high survivor strategy at high elevations.
7This is the first paper to show a shift in avian life-history strategies along an elevational gradient (in both genders, of multiple age classes) when region (latitude) and phylogenetic histories are controlled for.
Species that breed along steep elevation gradients experience very different conditions over relatively short spatial scales. With increasing elevation, temperature and growing season length decrease, storms become more frequent, snow cover persists for longer periods, plant productivity is lower, and there are seasonal delays in insect and fruit emergence (Körner 1999; Hegelbach 2001; Bears, Smith & Wingfield 2003). Vertebrates vary in their responses to elevational clines. Tropical species are often restricted to narrow elevational bands that encompass conditions to which they are adapted (Ghalambor etal. 2006). Some species may breed over wide elevation ranges due to a despotic distribution: i.e. competitive individuals occupy the preferred elevations, while inferior competitors breed in suboptimal elevations (Pearson & Rohwer 1998; Martin & Wiebe 2004; Rohwer 2004). However, some species that breed over wide elevational ranges adopt elevation-specific life-history strategies to maximize their lifetime reproductive success (Dunmire 1960; Zammuto & Millar 1985), involving functional adaptations and trade-offs to deal with differing environments. This latter group of species is ideal for comparative studies to determine life-history traits that may respond to selective pressure or are constrained among elevations within regions, as they avoid potentially confounding influences of different phylogenetic histories or geographical zones.
Birds at higher elevations typically breed for shorter periods and produce fewer broods of an equivalent or smaller clutch size than at lower elevations (Krementz & Handford 1984; Hamann etal. 1989; Badyaev 1997). The mechanisms orchestrating such shifts are unclear, however. An increase in spring photoperiod is the main cue that birds use to time the physiological, morphological, and behavioural changes required to initiate breeding (Hamner 1966; Farner & Lewis 1971). Eventually, birds become unresponsive to long days, reproduction ceases, gonads regress, and moulting begins (Farner etal. 1983; Nicholls, Goldsmith & Dawson 1988). However, since day length is the same among elevations at the same latitude, birds at higher elevations may modify their reproductive schedules in response to supplementary cues such as weather variables and food supplies, which differ across elevations. Alternatively, birds breeding at different elevations may modify the daylength to which they respond via local adaptation.
There is some evidence for trade-offs between reproduction and survival and/or offspring quality with elevation; birds at higher elevations that have lower seasonal production may survive longer (Sandercock etal. 2005a,b), produce higher quality offspring (Badyaev & Ghalambor 2001; Bears 2002), or have fewer parasitic infections (Stabler, Kitzmiller & Braun 1974; Bears 2004). In order to evaluate shifts in overall life-history strategies as opposed to elevation-specific variation in vital rates, there is a need to study multiple traits in a single, controlled system, enabling one to observe inter-related traits that co-vary with elevation. Sandercock etal. (2005a,b) showed the benefits of analyzing a suite of life-history traits; they showed that birds switched from a high-reproductive to a high-survival life-history strategy with increasing elevation. However, these authors compared congeneric grouse among elevations and could not control for phylogenetic or latitudinal differences; thus, it was unclear whether patterns observed were due to different phylogenetic histories, confounding regional and/or latitudinal differences, or breeding elevation.
We compared multiple interrelated life-history variables in a single songbird species, the dark-eyed junco (Junco hyemalis), at low- (1000 m; montane) and high- (2000 m; sub-alpine) elevation extremes of its breeding range in Jasper National Park, Alberta, Canada. We compared shifts in timing of growth and recrudescence of gonads, reproductive schedules, seasonal phenology, reproductive success, indicators of competitive status (arrival time, age, and body size), and local survival of dark-eyed juncos between the two elevations over a 5-year period. Over the 1000-m gradient, high-elevation habitats had a shorter period of food availability and a delayed, shortened period during which climatic conditions were favourable for breeding. We used this system to test the following hypotheses:
1Reproductive restriction hypothesis – birds in high- compared to low-elevation habitats should have a shorter annual breeding period, corresponding to a delay in the timing of gonad development in one or both sexes.
2Reproductive reduction hypothesis – high-elevation compared to low-elevation birds should produce fewer offspring that survive to fledging per season due to a shorter breeding period that restricts the number of potential broods. The shorter breeding season could be due to a shorter period of suitable weather conditions, available food, or predation levels conducive to successful breeding attempts; we assume some or all of these proximate variables may restrict when birds can breed. Since these variables co-vary with elevation in a predictable manner, we focus on elevation as the predictor variable encompassing all of these.
3Despotic distribution hypothesis – competitive exclusion forces less competitive birds to occupy high-elevation habitats, explaining the reduction in seasonal vital rates.
4Life-history trade-off hypothesis – benefits such as increased survival may compensate birds breeding at higher elevations that experience lower annual reproduction.
Dark-eyed juncos breed from sea level to the sub-alpine treeline in North America. They produce one to three clutches of two to five eggs, which are incubated by females, but both males and females feed and defend young (Nolan etal. 2002). Dark-eyed juncos show consistent natal philopatry and site fidelity in many regions and can survive for 3 to 11 years in the wild (Nolan etal. 2002). There is a well-established competitive hierarchy in dark-eyed juncos, where younger, smaller, and later arriving birds compete less successfully for preferred resources and breeding areas (Cristol, Nolan & Ketterson 1990; Grasso, Savalli & Mumme 1996). Hatchlings and young fledglings rely on insects and berries. Most dark-eyed juncos in our study region nest on the ground. Additional information on this species and study system are provided in Bears (2007) and Bears, Drever & Martin (2008).
field site description and procedures
Dark-eyed juncos were captured and monitored at four high and four low 50–70-ha sites from 2000 to 2005 in Jasper National Park (52°53′ N, 118°3′ W), Alberta. Sites are described elsewhere (Bears etal. 2003, 2008). Differences in weather conditions and food abundance (insect biomass) across the breeding season at each elevation are presented in Appendix S1.
Juncos were captured in mist nets using taped songs/calls of male juncos, marked with US Fish and Wildlife Service aluminum bands, given unique colour band combinations, and standard morphological measures taken (Bears etal. 2008). Juncos were aged as hatching year (HY), second year (SY), after second-year age (ASY), or unknown, according to Pyle (1997). We caught similar number of new (unbanded) juncos from each site during each sampling bout (two sites visited per day, each bout c. 4 days), allowing us to determine the proportion of birds in various reproductive stages at each elevation throughout the season. We observed activities at the nest, morphological features associated with reproduction, and other behavioural clues (e.g. delivery of insects to and from an area with a nest) in order to rank birds into the following breeding status categories: unpaired, pre-laying, laying, incubating, tending to nestlings, tending to fledglings, post-reproductive (male gonads recrudescing, female brood patch re-feathering, or late breeding season body moult). Sample sizes varied depending on whether measurements were collected via direct (i.e. capture) or indirect (i.e. observational) techniques and the number of years that the trait was measured. Territories of captured birds were mapped using global positioning system (GPS).
Surveys were conducted on each of the eight study sites each year and month during the field season. At sunrise, two observers walked transects stopping at every corner of a 100 × 100-m grid network. When a junco was observed, its banding status and identity were recorded.
reproductive timing and seasonal reproductive success
In 2000, surveys were conducted every 1–3 days, from 1–15 April, to determine arrival times in high- and low-elevation sites. In all survey years that spanned the entire breeding season (2000, 2001, 2004), banded birds were located in their territories weekly and observed to determine the temporal schedules of reproductive activities in high- and low-elevation sites. We relied heavily on behavioural observations of banded birds to determine their reproductive stages. It was not possible to use traditional nest-monitoring methods, since a large proportion of high-elevation nests were subterranean and inaccessible; juncos were found to be nesting up to 70 cm below ground in existing burrows and moss holes from roots of overturned trees (Bears 2002, 2007).
Weekly visits to territories of banded birds allowed us to determine the onset of breeding, hatching and fledging dates, initiation of subsequent nesting attempts, and cessation of breeding. We located hatchlings that were about to leave or had just left the nest, using cues from known pairs that were aggressive. Both parents made frequent feeding visits to each recent fledgling, which were typically located within c. 5–10 m2 of the nest site. We counted, and aged recent fledglings, defined here as young birds that had left the nest but could not yet fly. Fledglings at this stage were aged as 10–12 days according to Nolan etal. (2002).
We used observations of hatchlings and recent fledglings to determine the date of clutch initiation by backdating, using the estimated brood rearing period (10–12 days), the incubation period (12 ± 1 days for high, n = 4 nests and 12 ± 1 days for low, n = 10) and the elevation-specific clutch size (3·8 ± 0·8 eggs for high, n = 9 clutches and 3·5 ± 0·5 eggs for low, n = 15; 1 day per egg laid). Each estimate made using this method had a maximum error of ± 4–5 days. Eggs and hatchlings were not handled to minimize disturbance.
To estimate annual production, we repeated brood counts when fledglings reached 25–30 days of age and were capable of strong flight, but were still in family groups. We observed known families for 15–30 min until we identified every fledgling in the family. As there was more potential error associated with counts of older fledglings, we relied more on counts of recent fledglings in our interpretations. It was not possible to calculate life-history schedules for 2005, as the field season did not span the entire breeding period.
variation in reproductive development and termination
In 2004, we measured morphological indicators of reproductive stage to compare when high- and low- elevation populations initiated and terminated breeding capacities (i.e. grew or recrudesced reproductive features). When male songbirds become reproductively capable of breeding in the spring, the distal ends of the ductus deferens fill with new spermatozoa resulting in a swelling of the cloaca referred to as the cloacal protuberance (CP), which is the external male genitalia (King 1981). CP dimensions vary in size over the breeding season, and can be used as a measure of reproductive capacity (Lombardo 2001). A CP of c. one-half of the maximum size is considered functional for reproduction (Gwinner 1986). For males, we measured the cloacal length, width, and volume (King 1981; Kempenaers etal. 1999), which gave comparable results. Thus, only CP widths were presented, as in Deviche, Wingfield & Sharp (2000). Since CP size is not a reliable indicator of reproductive stage in females, we examined brood patch area as an indicator of reproductive stage; brood patches develop in response to increasing oestrogen and prolactin, enabling reproduction in females (Baily 1952). The higher the brood patch score, the more advanced a female was in her reproductive cycle.
We determined the termination of seasonal reproduction in males using primary moult score and decreasing cloacal width. Since moulting indicates that juncos have completed reproductive activities (Ginn & Melville 1983), we examined the schedules of pre-basic primary wing feather moult of males throughout the 2004 season at both elevations. We gave a value of 0 to old, unmoulted feathers, 1–4 to missing feathers and feathers growing in, and 5 to a completely grown, new feather (Ginn & Melville 1983). Moult scores were calculated as the mean score of all flight feathers (3 tertials, 10 primaries, 6 secondaries) on the right wing of birds captured at each time interval.
apparent survival analysis
Due to time restrictions, search time per transect per site varied during some months; but were always equivalent between elevations during the same periods (the comparison of interest here). We used the program mark with the Cormack–Jolly–Seber models for live encounter data (Cormack 1964; Jolly 1965; Seber 1965; White & Burnham 1999) to conduct a survival analysis, using 140 high- and 153 low-elevation SY and ASY banded males. Data from SY and ASY males were analysed for 15 occasions with unequal time intervals. We considered 17 models incorporating elevation as an attribute group and time effects in apparent survival (ϕ) and re-sighting (P) probabilities, with apparent survival standardized to 1-month intervals during the breeding season and over winter as one pooled interval. Interval lengths in months, starting in May 2000, were 1, 1, 1, 9 (August 2000 to May 2001), 1, 1, 1, 32 (August 2001 to April 2004), 1, 1, 1, 1, 10 (August 2004 to June 2005), and 1 month(s). For instance, when birds were last recaptured in August and then captured again the April of the following year, the interval length was longer, as there were longer gaps between recapture attempts during the winter months, whereas the interval was 1 between consecutive months during the same summer. A thorough explanation on integrating variable gap lengths into mark models can be found in Cooch & White (2006). We used information-theoretic procedures with Akaike information criterion (AICc) for model selection and evaluated goodness-of-fit of the global model [ϕ(elevation*t) P(elevation*t)] with the procedures in program release (Burnham etal. 1987; Burnham & Anderson 2002).
Data are presented as means ± 1 standard error (n, number birds). We pooled data across study sites by elevation to increase power, given a lack of significant interaction terms in initial nested anova (two-way analyses of variance) with site nested by elevation. We used a balanced design in site sampling, so means from pooled sites within an elevation would not be disproportionately affected by a potentially anomalous site. We tested data using two-way anova followed by Tukey's post-hoc tests. F values are presented as F[df(effect)/df(error)]. For female brood patch scores, we used Friedman's anova on ranks, followed by Holm–Sidak post-hoc tests as sample sizes were small, and data were non-normally distributed, with unequal variances (Sokal & Rohlf 1995). Survival modelling was performed using program mark (White & Burnham 1999), while other tests (e.g. t-tests) were performed in spss 11 (Statistical Package for the Social Sciences). Tests were two tailed and alpha (α) was 0·05 (Bonferroni adjusted as indicated). Partial eta-squared values (eta2), which represented the proportion of variance of the dependent variable explained by an independent variable, were included as a standard post-hoc measure of effect size. Strengths of eta2 values were interpreted according to Cohen (1988) where 0·01 = small effect, 0·06 = moderate effect, and 0·14 = large effect. Statistical power was > 0·90 unless otherwise indicated.
variation in settlement and competitive status by elevation
Dark-eyed juncos arrived between 2 and 7 April 2000 at high- and low-elevation breeding sites. The ratios of SY:ASY males in high-elevation (36:54) and low-elevation (43:53) habitats did not differ (Fisher's exact test, P = 0·30). The numbers of SY:ASY females in high-elevation (7:8) vs. low-elevation (8:8) habitats also did not differ (Fisher's exact test, P = 0·71). Thus, neither age structure nor arrival date varied by elevation for either sex.
variation in reproductive development and termination
Males: a two-way repeated measures anova on CP width in males revealed a large interaction effect between two elevations and 5 months (F4,146 = 2·22, P = 0·001, eta2 = 0·72, Fig. 1a). A Tukey's post-hoc test, revealed differences in CP widths between elevations at the first, second, and fourth capture intervals. High-elevation males had smaller CP widths compared to low-elevation males early in the breeding season until late May, and their CP widths decreased again by 1 July Low-elevation males, on the other hand, developed functional gonads before the beginning of the monitoring period and retained high CP volumes for 1 month longer. Mean CP widths of SY birds were slightly lower than those of ASY birds at every sample interval at both elevations, but this difference was not significant within either elevation (F1,146 = 1·00, P > 0·32). CP widths did not differ between elevations when categorized by life-history stages, with the exception that males with fledglings had smaller CP widths in high- vs. low-elevation sites (F4,228 = 6·32, P = 0·02; Fig. 1b). Females: elevation had a significant impact on timing of brood patch development (Friedman's test, = 54·79, P = 0·001, n = 64 females). Females in low-elevation habitat had more advanced brood patch development until mid-June compared to high-elevation females (Holm Sidak post-hoc, P < 0·05, Fig. 2).
termination of reproduction
In males, primary moult scores (indicators of reproductive cessation) did not differ between high- and low-elevation males, but variation was high, and high-elevation males achieved higher moult scores by the final sampling period on 20 August (F4,76 = 3·60, P = 0·01; Fig. 3). In females, there was no difference in the timing of brood patch re-feathering between elevations (Holm Sidak post-hoc, P > 0·05; Fig. 2).
variation in egg-laying schedules (n = 3 years)
The date by which the first 10% of birds breeding at high elevations initiated egg-laying was 22 May to 5 June, and ranged from 32–41 days later across years compared to when the first 10% of birds began breeding within low elevations (23–30 April; Fig. 4). In all years, birds in high-elevation habitats also ended their breeding seasons earlier than in low elevations, as 90% of broods were completed 15–20 days earlier. Although median brood initiation occurred at similar times between elevations, breeding seasons lasted only 38–43 days at high elevations compared to 95–99 days at low elevations. Due to this large difference in the breeding season length, high-elevation birds never had more than one brood per season while low-elevation pairs frequently produced two broods (Table 1).
Table 1. Summary of fecundity-related vital rates and fledgling condition for dark-eyed junco populations breeding in high- (2000 m a.s.l.) and low- (1000 m a.s.l.) elevations
High mean (± 1SE)
Low mean (± 1SE)
Significance t-test unless otherwise specified
= mean number of successful broods per year × mean number young fledglings per brood (ca. 11 days).
= mean number successful broods per year × mean number young per brood surviving to 25 days.
No. of broods per year (range successful broods; n = females)
0·75 (± 0·2)
1·55 (± 0·5)
t = 3·42 (d.f. = 78)
(0–1; n = 32)
(0–3; n = 48)
P < 0·01
No. of young fledglings per brood (surviving to 11 days)
2·7 (± 0·2)
2·9 (± 0·2)
t = 0·71
(n = 40)
(n = 45)
d.f. = 83, P = 0·48
No. of young fledglings per female, per season (including all broods)a
2·0 (± 0·5)
4·5 (± 1·0)
Mean difference = 2·5
(n = 32 females)
(n = 48 females)
(66% > at low elevation)
No. of fledglings per brood surviving to 25 days
2·5 (± 0·3)
2·3 (± 0·5)
t = 2·4
(n = 20)
(n = 20 broods)
d.f. = 38, P = 0·73
Survival rate, no. (percentage) of hatchlings surviving to 25 daysb
1·9 out of 2·0
3·6 out of 4·5 (80·1%)
d.f. = 1 P = 0·05
No. of young surviving to 25 days per female per season
1·4 ± (0·7)
3·6 (± 1·2)
Difference = 2·2
(39% > at low elevation)
Fledgling mass (g)
20·0 (± 0·8)
17·8 (± 1·2)
t = 1·32
(n = 11 broods)
(n = 18 broods)
d.f. = 27 , P = 0·20
Fledgling intrafurcular fat score (1–5)
4·2 (± 0·3)
1·7 (± 1·0)
t = 1·91
(n = 11 broods)
(n = 18 broods)
d.f. = 27, P = 0·07
variation in seasonal reproductive performance
A summary of vital rates related to fecundity are presented in Table 1. Although high-elevation females produced fewer broods per year, the mean number of young per brood, as counted as young fledglings, did not differ between elevations. The average survival rate of young birds until 25–30 days of age was 93% in high- vs. 79% in low-elevation habitats. The average number of fledglings at 25–30 days of age observed with parents at high- and low- elevations (per brood) did not differ. Thus, the average high-elevation junco produced fewer successful hatchlings per year overall, and the disparity was mainly due to the large differences in the number of broods attempted per season.
variation in offspring quality
The average fledgling weight was determined for broods where ≥ 2 individuals/brood were caught. By 25 days of age, fledglings at high elevations were c. 11% heavier and the average fat score in fledglings was more than double fledglings at low elevations. Thus, juncos at high elevations produced fewer, yet heavier offspring with greater fat reserves (i.e. potentially in better condition).
variation in apparent survival
Goodness-of-fit tests conducted in release in program mark suggested no correction for over-dispersion was required for fit to the global model (χ2 = 28·0, d.f. = 39, P = 0·90). Model selection results provided overwhelming support (AICc weighting of 0·99) for the top model [ϕ(elevation + t) p(elevation)], indicating that variation in survival probability (ϕ) was best explained by differences in elevation (Table 2). The estimated difference on the logit scale between the high- and low-elevation populations was –1·42 ± 0·37 SE, 95% CI –2·14 to –0·69, demonstrating higher survival estimates for males in high elevations at every time interval examined (Fig. 5). Survival of hatchlings and females could not be modelled due to sparseness of data. However, in 2001, we re-captured or re-sighted 20% of 10 birds from high- and 25% of 20 birds from low elevations that were banded as hatchlings in 2000, providing qualitative evidence for natal philopatry at both elevations.
Table 2. Model rankings and criteria from the program mark for the predominant factors tested to explain male dark-eyed junco monthly apparent survival (ϕ) and re-sighting probabilities (P) from 2000–2005 in high- and low-elevation sites in Jasper National Park, Canada
No. of parameters
Abbreviations are as follows: Akaike's information criterion (AIC), the difference between AICc values between models (ΔAICc), and AICc weights (probability that the model is the best among set of candidate models). The period symbol (.) means that the parameter was constant in the model. One top model provides overwhelming support for elevation in explaining variation in survival. The global model is in bold.
[ϕ(elevation + t) p(elevation)]
[ϕ (.) p(.)]
Overall, we found that most vital rates varied with elevation. Our data supported a shift to a slower life-history strategy with increasing elevation. We demonstrated four main results. First, the reproductive restriction hypothesis was supported, since birds in high-elevation habitat reduced the length of their reproductive season relative to low-elevation birds. The compressed breeding schedule corresponded to a delay in growth and earlier termination of reproductive structures in both sexes. Second, the reproductive reduction hypothesis was supported, as high-elevation birds produced fewer broods and offspring per season compared to low-elevation birds. This trend was mainly driven by the fact that high-elevation birds produced a maximum of one brood per season while low-elevation pairs were multi-brooded. Despite a higher percentage of fledglings surviving from hatchling age to independent flight (25–30 days of age) at high elevations, this did not compensate for the 66% higher number of offspring produced at low elevations within a season. Third, we failed to find evidence for the despotic distribution hypothesis; high-elevation populations were not composed of a higher proportion of less-competitive classes of birds. Fourth, the trade-off hypothesis was supported, as survival in high-elevation males was higher than at low-elevations. Despite high-elevation pairs having far fewer offspring per season, high-elevation hatchlings had a slightly greater survival rate early in life and higher intrafurcular fat scores before migration. Taken together, we showed a strong influence of elevation on the life-history strategies of populations similar to those seen in interspecific comparisons.
Our first finding was that high-elevation juncos had a compressed egg-laying period relative to low-elevation juncos. It is common for avian species to limit reproductive activity to the time of year when reproductive attempts are more likely to succeed (Perrins 1970; Wingfield 1983). Egg production and offspring rearing are costly in terms of energy, nutrients, and future fitness and demand a high proportion of a bird's daily energy budget (Monaghan, Nager & Houston 1998; Stevenson & Bryant 2000). Thus, dark-eyed juncos from high-elevations may be energetically constrained to breed for less than half as long as those from low-elevations because of additional thermoregulatory costs and/or a later timing of food emergence (especially of insects and berries eaten by young birds, Appendix S1).
Constraints on high-elevation breeding times could be due to multiple factors. Colder temperatures experienced at higher elevations are associated with a later reproductive onset, resulting in shorter breeding seasons, fewer broods, and a decrease in seasonal reproductive success (e.g. Stevenson & Bryan 2000; Naef-Daenzer etal. 2004; Weggler 2006). Persistent snow cover in spring can delay reproduction in ground nesting birds, resulting in fewer breeding attempts (Hendricks 2003). Severe storms in the spring can also delay egg laying and increase reproductive failure (Coulter & Bryan 1995; Martin & Wiebe 2004). Finally, delays in food emergence were correlated with delayed egg laying and lower seasonal productivity in great and blue tits (Parus major and Parus caeruleus; Svensson & Nilsson 1995; Thomas etal. 2001) and Eurasian dippers (Cinclus cinclus; Hegelbach 2001). Since these variables shift collectively with increasing elevation (Appendix S1), one might expect the negative correlation between seasonal reproductive success and breeding elevation that we and others documented (Krementz & Handford 1984; Hamann etal. 1989; Badyaev 1997).
High-elevation juncos terminated breeding earlier than low-elevation juncos, with no second broods observed at high elevations. The first clutch was likely initiated late enough that high-elevation birds would not have sufficient time to raise a second brood before the onset of their energetically expensive pre-basic moult began (Lindström, Visser & Daan 1993; Schieltz & Murphy 1995; Dawson etal. 2000). In contrast, male song sparrows (Melospiza melodia) breeding at 1210 m initiated primary moult later compared to birds in lower elevation sites (270 m, Perfito etal. 2004). Song sparrows breeding at 1210 m, however, may have encountered fewer late season constraints compared to dark-eyed juncos, as the upper elevation used in the Perfito etal. (2004) study is roughly equivalent to the low-elevation sites used in the present study.
Concomitant with the different egg-laying schedules, males and females differed in the phenology of growth and termination of reproductive features between elevations. In high- compared to low-elevations, males enlarged their CP widths later and began cloacal recrudescence and primary moult earlier, while females formed full brood patches later in the breeding season. When we examined male CP width by breeding stage, we found that high-elevation males regressed their CP while attending their first brood of fledglings, as they no longer required active testes for a second brood. Likewise, females had more extensively re-feathered brood patches by the last sampling interval in high- compared to low-elevation habitats. Thus, our data showed that high- and low-elevation birds become physiologically capable of reproduction at different times and may be confined to reproduce within different appropriate breeding windows (Thomas etal. 2001).
The differences in timing of reproduction associated with breeding between elevations raises questions about the cues used to initiate the physiological changes that precede clutch initiation. Daylength is normally the primary cue to initiate maturation of reproductive structures and subsequent moult (Nicholls etal. 1988; Wingfield & Kenagy 1991; Ball 1993), yet daylength is the same between elevations in our system. Thus, juncos may rely more heavily on supplementary cues (e.g. temperature, snow cover, food) or respond to different critical day lengths (e.g. Silverin, Massa &Stokkan 1993; Hahn 1998) to alter their reproductive schedules. There is evidence from a common garden experiment that birds taken from populations used in the present study respond to different critical daylengths and use supplementary cues for timing reproduction (i.e. populations from the two elevations have different reaction norms; Bears 2007).
Our second finding was that, due to the shorter breeding period in high-elevation habitats, high-elevation juncos produced fewer than half the offspring per season as their low-elevation conspecifics, mainly due to high-elevation birds producing less than half the number of broods. High-elevation birds did not compensate for having fewer broods by producing more chicks per brood, but did produce broods of similar sizes as low-elevation birds. Other intra- and interspecific studies on clutch size and fledgling production in songbirds have also demonstrated that clutch size remained constant (Hamann etal. 1989) or decreased with increasing breeding elevation (Krementz & Handford 1984; Badyaev 1997; Fargallo 2004; but see Purcell 2006).
Our third finding was that, although high-elevation birds had a lower seasonal reproductive success, they were not composed of inferior competitors (smaller, younger, later arriving individuals: Cristol etal. 1990; Grasso etal. 1996). Male black redstarts (Phoenicurus ochruros) had a similar age structure and pattern of breeding philopatry at high and low elevations (Köllinsky & Landmann 1996). Similarly, Widmer & Biebach (2001) found no difference in body sizes and age ratios of garden warblers (Sylvia borin) at the extremes of a 1000-m elevation gradient in Central Europe. However, male hermit (Dendroica occidentalis) and Townsend's warblers (Dendroica townsendi) arrived later at high-elevation breeding sites after spring migration, and yearlings were displaced into the higher-elevation habitats (Pearson & Rohwer 1998; Rohwer 2004). The greater fat reserves in high-elevation fledglings could be due to greater synchronicity between breeding and food supply peaks, less competition for food, a greater parental effort, a longer post-fledgling care period or a combination of these (Badyaev & Ghalambor 2001). Fledglings with greater fat reserves also survive better in the fall (Sullivan 1989; Leary, Sullivan & Hillgarth 1999; Oddie 2000).
Our fourth finding was that adult males in high-elevation habitats had higher survival rates. The differences revealed in survival models could also be due to higher emigration rates from low-elevation sites, as emigration and mortality cannot be differentiated in these models. If elevation influences true survival as the top model suggests, high-elevation birds may trade lower current reproduction for longer lives and higher quality offspring may compensate for their seasonal reproductive deficit on a lifetime basis if they and their progeny are collectively able to breed for more seasons. An inverse correlation between present and future reproduction and survival has been demonstrated in a wide range of vertebrates and invertebrates (Reznick 1985; Morton etal. 2004; Sandercock etal. 2005a; Parejo & Danchin 2006).
Multiple scenarios could explain a trade-off between reproduction and survival in birds. One scenario is that the high workload during reproduction compromises resistance to parasites and reduces survival (Stjernman, Raberg & Nilsson 2004). In support of this idea, we found lower incidences of blood parasites in high- compared to low-elevation adults and fledglings in this system (Bears 2004). Another scenario is that, if low-elevation birds shunt energy into reproduction for an extended period, adults and fledglings may end up at a lower nutritional state by the end of the season, decreasing their chances of over-winter survival (Badyaev 1997; Badyaev & Ghalambor 2001). Indeed, high-elevation birds produced fewer offspring but those offspring had more fat reserves. Earlier, we also showed that high-elevation male and female juncos experienced lower levels of physiological stress in response to noxious stimuli relative to low-elevation birds, which may protect them from stress-related illnesses (Bears etal. 2003). Predation of adults or fledglings may be greater at low elevations, such that these birds must produce as many young as possible early in life, given their lower survival within or between seasons. Indeed, proportionally more birds counted as hatchlings at low-elevations were gone during the second 25–30 day count (likely because they were depredated), although low-elevation birds still ended up with far more fledglings per pair than high-elevation birds. We did not measure predation rates after 25–30 days of age, but predation pressures in the closely related yellow-eyed junco (J. phaenotus) were less important once birds can fly (Sullivan 1989).
In conclusion, a number of inter-related life-history traits changed with elevation in a ground-nesting songbird. High-elevation dark-eyed juncos had a compressed egg-laying schedule and had fewer broods and offspring per season compared to low-elevation birds. Populations were not explained by despotic distributions; high-elevation populations were not composed of inferior birds. Instead, a shift from a high-reproductive to a high-survivor life-history strategy may occur with elevation. Our results fit the life-history shift expected with breeding elevation shown in an interspecific comparison (Sandercock etal. 2005a,b). This paper also suggests the exciting possibility that sufficient selective pressure for different phenotypes and life-history strategies exist over some elevation gradients, potentially promoting population differentiation necessary for speciation over a small spatial gradient (Bears etal. 2008).
Funding: Alberta Conservation Association Grant, Natural Sciences and Engineering Council of Canada (NSERC) PGS-D scholarship, and University of British Columbia Paetzold Fellowship to H. Bears, NSERC Discovery Grants to K. Martin, J.N.M Smith and Environment Canada support to K. Martin. G. Brown, S. Lord, and K. Keir provided field assistance. P. Arcese, J. N.M. Smith, D.E. Irwin, P.M. Schulte, M. C. Drever, B.K. Sandercock and L. J. Evans Ogden provided suggestions. K. Leinweber provided graphical aid. Experiments were carried out with appropriate permits (UBC Animal Care: #A04-0018; Jasper National Park: # 518, 038, 008, 005; Canadian Wildlife Service: CWS04-A001; Environment Canada: # 2000/067, WSA-6/00).