Abstract
Small-bodied birds engaging in incubation by a single sex experience a tradeoff between incubating to create a buffered thermal environment for their eggs and foraging to meet their own energetic requirements. This tradeoff is intensified in alpine environments, which are characterized by cold and variable conditions. We monitored the incubation rhythms of alpine Horned Larks (Eremophila alpestris) in British Columbia, Canada, across different annual thermal regimes (2005: moderate; 2006: warm; 2010: cold overnight; 2011: cold during the day). In this species, females alone incubated and left their nests to forage at dawn, following 7 hr of nighttime incubation in near-freezing conditions. However, with early morning ambient temperatures still <5°C, this placed embryos at high risk of chilling during incubation recesses. Focusing on behavioral decisions made by females at dawn (06:00–08:00 hours), we examined relationships between incubation rhythms and ambient temperature among years for evidence of variable responses to temperature. In all years, females spent more time off the nest at dawn in warmer temperatures, but in 2010, which was colder overnight, the slope of the line relating attentiveness to ambient temperature was steeper, indicating that females left their nests at colder temperatures compared with other years. In 2010 females also took shorter recesses at cold temperatures. Hatching success remained high in 2010 relative to warm or moderate years; however, overwinter survival of females declined to 48% from 2010 to 2011 compared with 72% in earlier years. When faced with exceptional thermal constraints, alpine Horned Larks made behavioral adjustments to their incubation rhythms and were able to maintain fecundity. However, potential survival costs to females implies a shift in balance of the parent–offspring tradeoff, revealing limits to coping mechanisms of alpine-breeding Horned Larks.
Résumé
Les oiseaux de petite taille dont un seul sexe s'occupe de l'incubation doivent faire un compromis entre l'incubation, pour créer un environnement thermique tamponné pour leurs œufs, et l'alimentation, pour répondre à leurs propres besoins énergétiques. Ce compromis s'intensifie dans les milieux alpins, caractérisés par le froid et des conditions variables. Nous avons suivi les rythmes d'incubation d'Eremophila alpestris en milieu alpin en Colombie-Britannique, au Canada, à travers différents régimes thermiques annuels (2005: modérée; 2006: chaud; 2010: froid la nuit; 2011: froid le jour). Chez cette espèce, les femelles incubaient seules et quittaient leur nid pour s'alimenter à l'aube, à la suite de 7 h d'incubation pendant la nuit dans des conditions près du point de congélation. Cependant, avec des températures ambiantes matinales encore sous les 5°C, cela a rendu les embryons à haut risque de refroidissement au cours des pauses d'incubation. En se concentrant sur les décisions comportementales des femelles à l'aube (0600–0800 heures), nous avons examiné les relations entre les rythmes d'incubation et la température ambiante entre les années pour des preuves de réponses variables à la température. À toutes les années, les femelles passaient plus de temps hors du nid à des températures plus chaudes à l'aube, mais en 2010, qui était plus froide la nuit, la pente de la ligne reliant la vigilance à la température ambiante était plus abrupte, ce qui indique que les femelles quittaient leur nid à des températures plus froides comparativement aux autres années. En 2010, les femelles ont également pris de plus courtes pauses à des températures froides. Le succès d'éclosion est demeuré élevé en 2010 comparativement aux années chaudes ou modérées; toutefois, la survie hivernale des femelles a chuté à 48 % de 2010 à 2011 comparativement à 72% les années précédentes. Lorsqu'elles doivent faire face à des contraintes thermiques exceptionnelles, les femelles d'E. alpestris en milieu alpin ont ajusté leur comportement d'incubation et ont réussi à maintenir leur fécondité. Toutefois, des coûts de survie potentiels pour les femelles impliquent un changement dans l'équilibre du compromis parent–progéniture, ce qui révèle des limites aux mécanismes d'adaptation d'E. alpestris nichant en milieu alpin.
Mots-clés: température ambiante, variation annuelle, vigilance, Eremophila alpestris, haute altitude, compromis parent-progéniture, survie
Introduction
Incubation is an energetically expensive phase of avian reproduction (Tatner and Bryant 1993, Williams 1996), during which parents must expend energy to create a buffered thermal environment for their eggs while also meeting their own somatic needs (White and Kinney 1974, Conway and Martin 2000a). In small-bodied species exhibiting single-sex intermittent incubation, this can lead to a tradeoff for the parent between incubating to ensure embryo viability and foraging for self-maintenance (Williams 1996, Conway and Martin 2000a, 2000b). This tradeoff is intensified in cold environments such as alpine habitats because as ambient temperatures approach extreme highs or lows, both the rate of egg warming or cooling and the metabolic rate of the parent increase (Drent 1975, Haftorn 1988, Martin and Wiebe 2004).
Avian embryos require a particular range of temperatures for normal development (Webb 1987, Haftorn 1988), and in temperate ecosystems this range is typically higher than ambient conditions. Periodic cooling during incubation can reduce embryonic growth efficiency and produce lower-quality nestlings (Olson et al. 2006, Ardia et al. 2010). To ensure high clutch survival, parents should increase attentiveness at low temperatures. However, the energetic needs of parents also change with ambient conditions, such that at low temperatures parents may need to reduce nest attentiveness and feed or forage more to compensate for their increased metabolic requirements. The conflicting needs of parents and embryos, and the fact that foraging and incubating are mutually exclusive, means that parents may need to adjust their behavior to maximize fitness (Jones 1989).
Given the potential conflicting effects on both embryos and parents, ambient temperature should influence incubation behavior. A relationship between incubation rhythms and temperature has been demonstrated in several species across various habitats (Conway and Martin 2000a, Londono et al. 2008, Camfield and Martin 2009, Kovarik et al. 2009). Conway and Martin (2000a) proposed a nonlinear framework for how recess and on-bout duration of small-bodied intermittent incubators should change with ambient temperature, finding support for their hypothesis in observations of Orange-crowned Warblers (Oreothlypis celata) in Arizona, USA. However, no relationship between recess duration and ambient temperature was observed for alpine Horned Larks (Eremophila alpestris) incubating in British Columbia, Canada, or alpine Meadow Pipits (Anthus pratensis) in the Czech Republic (Camfield and Martin 2009, Kovarik et al. 2009). These inconsistent relationships between incubation rhythms and ambient temperature point to variability in the pattern across species, habitats, and populations. Finer-scale examination of the relationship between attentiveness and temperature may reveal sources of variability.
How an incubating bird adjusts its behavior in response to temperature should depend on the environment in which it is incubating. In the aforementioned studies, alpine-breeding Horned Larks and Meadow Pipits experienced much colder and more variable conditions than Orange-crowned Warblers in Arizona. Different environments could account for the conflicting results reported in these studies, as the response to temperature may depend on the range of temperatures to which a bird is exposed. Incubation behavior changes as adult energetic constraints are alleviated or heightened by experimental heating or cooling. Incubating Pectoral Sandpipers (Calidris melanotos) and Tree Swallows (Tachycineta bicolor) increased nest attentiveness in response to nest heating (Creswell et al. 2004, Ardia et al. 2009). Tree Swallows also responded to experimental cooling by reducing incubation time, which resulted in lower egg temperatures (Ardia et al. 2010). Similarly, Zebra Finches (Taeniopygia guttata) reduced egg temperatures when they incubated under colder conditions (Nord et al. 2010). If incubating birds are exposed to different thermal regimes or weather, particularly in environments that regularly experience wide-ranging temperatures, parents may adjust how they manage their time as the temperature changes.
Our objective was to determine whether alpine-breeding Horned Larks altered their incubation behavior in response to thermal conditions. Passerine foraging is often constrained by available daylight, forcing incubating parents to fast overnight when it is too dark to feed (Jenni and Jenni-Eiermann 1996). In the alpine environment, females experience >7 hr of near- or below-freezing conditions during these overnight fasting periods (Camfield and Martin 2009); thus, the energetic costs of behavioral decisions in the early morning are likely heightened. We focused on behavior in that early morning period and examined relationships between incubation attentiveness and ambient temperature across four years with distinct thermal regimes. We expected that the relationship between time on the nest (attentiveness) and ambient temperature in the early morning would differ among years, reflecting variation in ambient conditions. We predicted that in colder years (due to greater thermal stress overall), females should face stronger energetic tradeoffs between incubation and self-maintenance and would alter time on the nest to reflect greater investment in their offspring. We also examined temperature relationships with on-bout duration, recess duration, and recess frequency to investigate how larks adjusted attentiveness. Given current and anticipated increases in the variability of thermal regimes, especially at high elevations (IPCC 2013), we evaluated evidence for potential fitness consequences (fecundity or annual survival) stemming from behavioral changes in response to more extreme ambient conditions, expecting that colder regimes and altered behavioral strategies could impact nesting success and/or adult survival.
Methods
Study Species and Study Site
Horned Larks (Eremophila alpestris) are small-bodied (28–40 g), ground-nesting passerines exhibiting female-only intermittent incubation with no evidence of male incubation feeding (Beason 1995). We studied a population of Horned Larks on Hudson Bay Mountain near Smithers, British Columbia, Canada (52°N, 127°W), during the 2005, 2006, 2010, and 2011 breeding seasons. Data were collected in 2005 and 2006 by A. F. Camfield and in 2010 and 2011 by E. C. MacDonald (Camfield 2008, MacDonald 2012). Birds breeding at this site face fluctuating daily temperatures, dropping to near or below freezing every night and sometimes reaching over 40°C during the day (Camfield and Martin 2009). High winds, heavy fog, and storms are common during the breeding season. The structurally simple tundra provides limited protection for nests (Martin and Wiebe 2004). Timing of snowmelt varies annually, and snow persists throughout most of June. Warmer springs usually result in earlier arrival at breeding areas and earlier clutch initiation (Murphy-Klassen et al. 2005, Dunn and Winkler 1999). Mean clutch initiation date for Horned Larks was May 31, with date of first egg varying from May 17 to June 13 among years (Camfield et al. 2010, MacDonald 2012).
Available Daylight and Onset of the Active Day
Given the latitude of our field site, days are long during the breeding season. At summer solstice there is ∼17.5 hr of daylight, with an additional 2 hr of civil twilight (National Research Council of Canada, http://www.nrc-cnrc.gc.ca/eng/services/sunrise/), therefore larks in this population are probably not limited by daylight for foraging. Previous work has designated 06:00–20:00 hours as the active day for Horned Larks at this site (Camfield and Martin 2009, MacDonald et al. 2013), thus we have selected 06:00–08:00 hours as the first two hours of the active day.
Data Collection
Each year, between May and early August, we located Horned Lark nests by searching known territories or appropriate habitat and by observing females approaching or leaving their nest during nest building, egg laying, and incubation. We monitored nests every 3–5 days, more frequently near expected hatching and fledging dates, to record nesting status. If nests were located after the onset of incubation, we calculated the clutch initiation date by backdating from hatch using the following intervals: egg laying = 1 egg laid per day (thus, clutch size = duration of laying period), incubation = 12 days. To estimate fecundity and survival, we banded and individually color-marked females for resighting (less effort in 2010 and 2011) using a bow trap to capture females at the nest. Each year we conducted systematic surveys of the study area to determine which banded individuals had returned (see Camfield et al. 2010 for further detail).
At each nest site, we measured percent overhead cover, nest orientation, and aspect of the landscape slope (0 to 360°). Percent cover was estimated following Nelson and Martin (1999) using a ball marked with dots in a 1 cm × 1 cm grid. Nests were constructed at the base of a clump of grass, and nest orientation was estimated by the compass bearing of a line from the center of the clump of grass to the center of the nest.
We used temperature data loggers (HOBO Pro Series, #H08-031-08, Onset Computer Corporation, Pocaset, Massachusetts, USA) to monitor incubation rhythms. A temperature logger consisted of a unit containing both the data logger and an internal temperature probe as well as an external temperature probe that extended from the unit by a cord. The unit measuring ambient temperature was placed within 2 m of the nest and the cord was extended from the unit such that the external temperature probe was inserted into the nest cup to measure nest temperature. These two probes took simultaneous recordings of ambient and nest temperatures (°C) every 30 s, 24 hr per day, until the nest hatched or failed. Data loggers did not appear to influence nest success, as the proportion of nests that hatched and failed to hatch did not differ among nests with or without HOBOs (χ21 = 0.002, n = 212, P = 0.97). We used nest temperature fluctuations to infer when the incubating female left and returned to her clutch. Behavioral observations at multiple nests over a range of temperatures and years verified that temperature fluctuations recorded by data loggers coincided with actual incubation recesses and on-bouts (Camfield 2008, MacDonald 2012). As Horned Larks are ground-nesting songbirds, ambient temperatures on the ground closely approximate temperatures experienced at the nest; therefore, we placed the unit measuring ambient temperature in areas with similar aspect and protection from the wind as the nest. The external probe did not touch the eggs, so we did not measure the actual egg/embryo temperatures.
A weather station at the field site recorded daily temperature throughout all breeding seasons, allowing us to characterize the thermal regime that Horned Larks experienced during each year of study.
Data Analysis
To classify years relative to one another in terms of thermal regime, we used a method of cumulative daily temperature deviations following Myers and Pitelka (1979). For all four years, we used daily (24 hr) temperature readings from the weather station throughout the breeding season (May 14 to July 22). We calculated mean daily ambient temperature for each day within each year as well as among years. The deviation from this daily mean within each year was calculated by subtracting the mean for each day within that year from the among-year mean for that day. We then summed this deviation across the breeding season within each year to come up with a cumulative deviation from the “normal” temperature regime for these four years. This allowed us to compare temperature regimes to assess variability in annual conditions experienced by breeding birds. We refined these comparisons by examining ambient temperature variation throughout the day. We divided the day into twelve 2-hr periods, and for each period we calculated mean ambient temperature (°C) at the nest in each year to see if annual thermal regimes varied due to differences at particular times of day. We compared mean ambient temperature at the nest among years using a Kruskal-Wallis test (nonparametric ANOVA) as the data failed to meet assumptions of parametric testing.
We used program Rhythm 1.0 (Cooper and Mills 2005) to select incubation recesses from nest temperature recordings. We considered an observation a recess if the nest temperature dropped by more than 3°C for at least 3 min (following Camfield and Martin 2009). We then visually inspected the data using Raven Pro 1.3 to confirm the recesses selected by Rhythm 1.0 and manually selected any recesses that Rhythm 1.0 had not chosen but were also incubation off-bouts (i.e. a steep temperature drop lasting slightly less than 3 min). From these data, we calculated initiation and duration of each incubation recess and on-bout. For each of the twelve 2-hr periods, we calculated the total number of minutes the female was on the nest, the number of recesses, mean recess and on-bout duration, and proportion of time spent on the nest. Incubation recesses and on-bouts were not necessarily bound by the 2-hr periods and thus some observations were >120 min. If an activity (on-bout or recess) extended across multiple 2-hr periods, we used total time for the activity when calculating period means. This is particularly relevant for on-bouts between 06:00 and 08:00 hours because overnight incubation (which typically began between 20:00 and 23:00 hours) sometimes continued into (and beyond) the 06:00–08:00 hours period; thus, some mean values included overnight on-bouts.
To assess the relationship between attentiveness and ambient temperature as well as variability in this relationship across years, we used linear mixed effects models with a Restricted Maximum Likelihood (REML) method of parameter estimation in the statistical package R (nlme, R Version 2.10.1; R Development Core Team 2009). Mixed effects models allowed nest to be treated as a random effect, avoiding pseudoreplication, as nests were sampled repeatedly over time (Pinheiro and Bates 2000). Nonlinear models may be the most appropriate way to view the relationship between incubation rhythms and temperature across a wide temperature gradient during the entire day (Conway and Martin 2000a, Camfield and Martin 2009). However, we chose a linear model because the temperatures experienced between 06:00 and 08:00 hours were within the temperature range where Camfield and Martin (2009) observed a linear relationship. Visual inspection of the data also suggested that a linear model was appropriate.
Total time on the nest (min) from 06:00 to 08:00 hours was modeled as the response variable, with ambient temperature at the nest as a fixed effect and nest as a random factor. We used nest ID rather than female ID as the random factor because not all females in the study were banded and 10 females were monitored twice among or within years. We performed our analyses with and without replicate females and found no difference in results. Consequently, we treated each nest as independent (following Camfield and Martin 2009). Year was also a fixed factor, and a temperature × year interaction was included to determine if the relationship between time on the nest and temperature varied among years. These models also controlled for several other factors that can influence songbird incubation behavior, including clutch size, date, incubation day, percent vegetation cover at the nest, nest orientation, aspect of the nest location, and the proportion of time spent in overnight incubation from 20:00 to 05:59 hours (included as fixed effects; Smith 1989, Wiebe and Martin 1998, Conway and Martin 2000b, Joyce et al. 2001, Wheelwright and Beagley 2005, Zimmerling and Ankney 2005, Kovarik et al. 2009). Vegetation cover and overnight attentiveness are proportion data and were therefore arcsine square-root transformed (Zar 1999). Nest orientation and aspect were assigned to one of four categories based on their deviation from north (N, E, S, or W).
To determine whether other incubation behaviors varied with temperature, we examined relationships for mean incubation on-bout duration, mean recess duration, and recess frequency with ambient temperature using the same fixed and random effects as above. On-bout and recess duration were log-transformed to achieve normality and modeled using the same framework as time on the nest. Recess frequency was modeled using a generalized linear mixed effects model with a Poisson distribution (O'Hara and Kotze 2010; lme4, R Version 2.10.1; R Development Core Team 2009).
To evaluate the potential for fitness consequences related to adjustments in incubation rhythms in response to thermal regimes, we present fecundity data for the four study years for which we present behavioral data. Estimates of vital rates for this population were published earlier (2003 to 2006; Camfield et al. 2010). In this study, we include data from three additional study years (2007 [A. F. Camfield] and 2010 and 2011 [E. C. MacDonald]). We used Program MARK to measure apparent annual survival of adult females (White and Burnham 1999). New detections of banded larks missed from the earlier study resulted in slightly different resighting and survival estimates compared with what was reported in Camfield et al. (2010). No banding occurred from 2007 to 2009 and there was no resighting effort in 2008 and 2009, therefore the period between 2006 and 2010 (representing four survival transitions) was measured as a single average. We compared estimated survival from 2010 to 2011 with the average over the earlier period (2003–2010) to examine potential effects of the cold 2010 thermal regime.
We used Akaike's Information Criterion adjusted for small samples (AICc; Burnham and Anderson 2002) to test whether apparent annual survival of females was lower between 2010 and 2011 compared with all other years. A bootstrap goodness-of-fit was used to examine the fit of the time-dependent model (ϕtpt) for female survival. The variance inflation factor (ĉ) was used to examine overdispersion in the data and was calculated by dividing the observed model ĉ by the mean ĉ of 1,000 simulations (Cooch and White 2013). The estimate of ĉ was also used to adjust AICc to quasi AICc (Burnham and Anderson 2002). In all instances, means are presented ± SE.
Results
Incubation rhythms were recorded at 86 nests throughout the four years of study (2005: 16 nests; 2006: 38 nests; 2010: 16 nests; 2011: 16 nests). From 06:00 to 08:00 hours, incubating females spent an average of 89.71 ± 0.66 min of the total 120 min (75% of their time) on the nest. On average (including some overnight incubation), on-bout duration was 57.72 ± 6.15 min and females took 4.13 ± 0.09 recesses, which lasted 7.68 ± 0.27 min each. From 06:00 to 08:00 hours, ambient temperature at the nest varied among years (Kruskal-Wallis: χ23 =103.9, P < 0.001); 2010 was on average >4°C colder than 2005 and 2006 and >2°C colder than 2011 (Table 1). In general, at dawn ambient temperature was only slightly warmer than overnight (Figure 1A), leaving eggs at a high risk of cooling. The proportion of time spent on the nest (attentiveness) varied with time of day (Figure 1B). Despite variable overnight temperatures (means ranging from 0.25 to 4.76°C), overnight attentiveness was consistently high among years (ranging from 93 to 95%), while attentiveness throughout the day varied (Figure 1B).
Annual variation in ambient temperatures (°C) experienced at Hudson Bay Mountain, British Columbia, Canada, at dawn (06:00–08:00 hours). Letters indicate statistical differences (independent 2-group Mann-Whitney U-tests). Means are presented ± SE.
Annual variation in ambient temperatures (°C) experienced at Hudson Bay Mountain, British Columbia, Canada, at dawn (06:00–08:00 hours). Letters indicate statistical differences (independent 2-group Mann-Whitney U-tests). Means are presented ± SE.
Mean (± SE) (A) ambient temperature (°C) and (B) attentiveness (proportion of 2-hr period spent on nest) of incubating Horned Larks in each 2-hr time period at Hudson Bay Mountain, British Columbia, Canada, during the four years of study.
Throughout our study, for the range of dates that we had data loggers in nests, sunrise varied from 04:46 to 05:08 hours and civil twilight from 03:45 to 04:17 hours (National Research Council of Canada, http://www.nrc-cnrc.gc.ca/eng/services/sunrise/). Therefore, at the onset of our 2-hr observation period, females would have had nearly an hour (at minimum) after sunrise, and up to nearly 2 hr (at minimum) of available daylight for foraging; however, during the previous 2-hr period (04:00–06:00 hours), larks maintained nighttime incubation rhythms rather than transitioning to daytime behavior, exhibiting upwards of 90% attentiveness (Figure 1B).
Annual Thermal Regimes
Each of the four study years represented a different thermal regime. Cumulative temperature deviations from the daily average across the four years revealed that 2005 was moderate, 2006 from early June onward was warmer than the other three years, and 2010 and 2011 were generally colder (Figure 2). Furthermore, the two colder years differed, with 2010 being colder overnight and early in the morning than the other three years, while 2011 had comparable overnight temperatures to 2005 and 2006 but did not warm up during the day and had lower midday temperatures than 2010 (Figure 1A).
Cumulative temperature deviations from the daily mean at Hudson Bay Mountain, British Columbia, Canada, during the four years of study. A zero baseline indicates an average year, positive values indicate a relatively warmer year, and negative values indicate a relatively colder year.
Early Morning Incubation Rhythms and Ambient Temperature
Time on the nest from 06:00 to 08:00 hours exhibited a negative linear relationship with ambient temperature; incubating Horned Larks decreased time on the nest as ambient temperature increased (Figure 3A, Table 2). This relationship differed in one year of the study, with 2010 exhibiting a significantly steeper slope than the other years (Table 2, Figure 3A). Time on the nest also varied with the proportion of time spent incubating overnight (as expected when measuring two periods close in time); birds spent more time on the nest from 06:00 to 08:00 hours when they spent more time incubating overnight (Table 2, Figure 4A), although, in general, time on the nest overnight did not vary among years (Figure 1B). The unique relationship in 2010 appeared to be driven by colder temperatures in that year; however, 2010 was still significantly different when the data were reanalyzed excluding the lowest temperatures (−4.78 to −2.40°C, accounting for 2.6% of observations, n = 531) not experienced in the other years (contrast comparing 2010 vs. 2005, 2006, and 2011: t439 = −2.17, P = 0.03).
Linear relationships of (A) total time on the nest (min), (B) mean on-bout duration (min), (C) mean recess duration (min), and (D) recess frequency with ambient temperature from 06:00–08:00 hours for incubating Horned Larks at Hudson Bay Mountain, British Columbia, Canada, during the four years of study. Statistical results associated with these relationships are presented in Table 2.
Results of general linear mixed effects models predicting total minutes on the nest, mean on-bout duration, and mean recess duration, and generalized linear mixed effects model for recess frequency of Horned Larks incubating from 06:00–08:00 hours at Hudson Bay Mountain, British Columbia, Canada (n = 545 2-hr periods), over four years. On-bout and recess duration were log-transformed and proportion of time spent in overnight incubation was arcsine square-root transformed to achieve normality. Significant results are in bold font, and marginally significant results (interactions with 0.05 < P ≥ 0.10) are underlined. When the temperature × year interaction was significant, temperature and year main effects were not considered (even when P < 0.05). Model factors with no significant results are excluded from this table for brevity, but can be viewed in table 2.1 of MacDonald (2012).
Results of general linear mixed effects models predicting total minutes on the nest, mean on-bout duration, and mean recess duration, and generalized linear mixed effects model for recess frequency of Horned Larks incubating from 06:00–08:00 hours at Hudson Bay Mountain, British Columbia, Canada (n = 545 2-hr periods), over four years. On-bout and recess duration were log-transformed and proportion of time spent in overnight incubation was arcsine square-root transformed to achieve normality. Significant results are in bold font, and marginally significant results (interactions with 0.05 < P ≥ 0.10) are underlined. When the temperature × year interaction was significant, temperature and year main effects were not considered (even when P < 0.05). Model factors with no significant results are excluded from this table for brevity, but can be viewed in table 2.1 of MacDonald (2012).
Relationships (mean ± SE) during the 06:00–08:00 hours period of (A) total time on the nest, (B) mean on-bout duration, (C) mean recess duration, and (D) recess frequency with proportion of time spent in overnight incubation (20:00–05:59 hours) across the four years of study for alpine Horned Larks at Hudson Bay Mountain, British Columbia, Canada. Statistical results associated with these relationships are presented in Table 2.
Similarly to time on the nest, mean on-bout duration decreased linearly with increasing ambient temperature; however, this relationship did not vary statistically across years (Table 2, Figure 3B). Mean on-bout duration also varied with Julian date, incubation day, and proportion of time spent incubating overnight. On-bout duration decreased as Julian date and incubation day increased, and on-bouts were longer when more time was spent in overnight incubation (Table 2, Figure 4B).
The relationship between mean recess duration and ambient temperature varied in one year of the study. Recess duration increased with increasing ambient temperature in 2010, but appeared to be unrelated to temperature in all other years (Table 2, Figure 3C). Recess duration also varied with time spent in overnight incubation, decreasing with increased time on the nest overnight (Table 2, Figure 4C). These relationships were still significant when the low outlier points clustered on the bottom left portion of the figure (which represented mean recess durations of 0 min, where a bird did not leave the nest during those 06:00–08:00 hours observation) were excluded from analysis (contrast comparing 2010 vs. 2005, 2006, and 2011: t424 = 2.42, P = 0.02). The steeper, positive slope in 2010 seemed to be driven by the lower temperatures in that year, as 2010 was no longer significantly different when reanalyzed without the coldest temperatures (contrast comparing 2010 vs. 2005, 2006, and 2011: t439 = −1.36, P = 0.17).
Birds took more frequent recesses as ambient temperature increased in the early morning period and this relationship was consistent across years (Table 2, Figure 3D). Recess frequency also displayed a positive relationship with Julian date and incubation day and a negative relationship with proportion of time spent in overnight incubation (Table 2, Figure 4D). Nest site characteristics showed no significant relationships with incubation rhythms and are excluded from Table 2 for brevity (see MacDonald 2012).
Life History Correlates
Nesting success and nest survival rates varied among the four years. Nest success was higher in warm and moderate years and declined in colder years, coinciding with higher nest predation in 2010 and 2011 and more nest abandonment in 2011 (Table 3). Overall, hatching success was high (93%) in the first three years of our study but declined to 81% in 2011 (Table 3; MacDonald et al. 2013), while fledging rates were consistently high among years. From 2003 to 2010, average apparent annual female survival was 0.72 ± 0.05, but dropped to 0.48 ± 0.18 from 2010 to 2011. Based on quasi AICc, the top model did not support a separate estimate for 2010–2011 (QAICc = 0.57 units higher than a model with constant survival from 2003 to 2010). However, with only one cold year, confidence intervals were wide and we had limited power to detect a difference based on AICc, despite a 24% drop in mean annual survival.
Vital rates of Horned Larks breeding on Hudson Bay Mountain, British Columbia, Canada, during four years for which we present data on dawn incubation rhythms. Results are presented as means ± SE or proportions (prop.), and sample sizes are in parentheses. Julian dates are based on 1 = 1 January. Estimates for each parameter are based on the number of nests within that year for which data were available and include first nests, renests, and second broods.
Vital rates of Horned Larks breeding on Hudson Bay Mountain, British Columbia, Canada, during four years for which we present data on dawn incubation rhythms. Results are presented as means ± SE or proportions (prop.), and sample sizes are in parentheses. Julian dates are based on 1 = 1 January. Estimates for each parameter are based on the number of nests within that year for which data were available and include first nests, renests, and second broods.
Discussion
Horned Larks breeding on Hudson Bay Mountain, British Columbia, Canada, are confronted with strong annual variation in their thermal environment. In this study we also observed interannual variation in incubation behavior at dawn, which was seemingly related to variable thermal regimes. In 2010, when overnight temperatures were much colder, females left the nest at dawn in colder temperatures and spent less time on the nest as temperatures warmed than in other years. Females also took shorter recesses at colder temperatures in 2010, while in other warm or average years recess duration was not related to temperature. Hatching rate remained high in 2010, thus embryo viability appeared to be unaffected by the increased time off the nest at cold temperatures. However, a decline in female survival between 2010 and 2011 indicates that there may have been fitness consequences associated with the observed adjustments to incubation behavior. Our findings suggest that Horned Larks use nuanced behavioral adjustments to cope with changing energetic demands, although there may be limits to such coping mechanisms. The energetic constraints imposed by an exceptional thermal regime may result in a shift in balance of the parent–offspring tradeoff, potentially leaving females to pay a fitness cost in terms of survival in order to maintain fecundity.
Life history correlates of females and/or offspring would provide evidence for whether our observed behavioral changes were the product of an energetic tradeoff. Energetic stressors or constraints influence the management of the parent–offspring tradeoff through changes to relative concentrations of hormones like corticosterone and prolactin (Angelier and Chastel 2009, Spée et al. 2010). Assuming that increased time off the nest reflects the female favoring her somatic needs over embryonic thermal requirements (Voss et al. 2006, Ardia et al. 2009), the behavioral difference observed in 2010 could indicate that incubating females shifted investment toward their self-maintenance during energetically stressful conditions. Low overnight temperatures during incubation lead to increased metabolic heat production in order to sustain suitable egg temperatures (Carey 2002, Nord et al. 2010). Females in 2010 maintained high attentiveness overnight and did not reduce their overnight incubation bout to compensate for increased energetic demands, but rather potentially accrued greater energetic deficits. Incubating females in that year may have had to spend more time off the nest at dawn to meet their heightened energetic requirements. In support of this explanation, Ardia et al. (2010) found that Tree Swallows spent less time in incubation when nests were cooled experimentally, and Bryan and Bryant (1999) found that Great Tits were able to allocate more time to incubation when nests were experimentally heated overnight. Rates of hatching and fledging were high in 2010 (consistent with the earlier and warmer years), therefore the extra time that females spent off the nest at cold temperatures did not result in a fecundity cost in the form of reduced embryo viability.
In 2011, the relationship between time spent incubating and ambient temperature at dawn did not vary; however, females took extended incubation recesses, with breaks from incubation lasting between 1 and 6 hr at a time (MacDonald et al. 2013). Extended recesses may represent an “emergency life history stage,” when an organism shifts investment from reproduction to survival in response to an environmental perturbation (Wingfield 2003). Such a dramatic behavioral response is in contrast to the nuanced shift in early morning behavior observed in 2010 and is particularly interesting when we compare costs to embryo viability, as hatching rate declined in 2011 but not in 2010. Extended recesses seem to represent a shift in the parent–offspring tradeoff, in which the parent puts its somatic needs above embryonic thermal requirements. However, in the absence of female survival data from 2011, we cannot assess whether annual survival changed in that year and whether the females also experienced a tradeoff.
Indirect fitness costs are more difficult to establish. We observed a decline of nearly 25% in female survival in the coldest years compared to earlier estimates (Camfield et al. 2010), which could be associated with an uncompensated energy deficit that accrued through the 2010 breeding season. We found no evidence for adult mortality during the 2010 nesting effort. There are limits to what can be deduced from one year of survival data; however, this apparent reduction in female survival suggests that 2010 was a particularly challenging year. High predation at our field site, especially in the two latter years of the study, made it impossible to investigate direct effects of behavior on individual nest success; however, overall hatching success declined in 2011 relative to the first three years of our study. These observations provide evidence that female larks in 2010 may have been forced to trade off survival for fecundity.
Despite adjustments to incubation rhythms in 2010, larks maintained high overall attentiveness in that year, particularly at very low temperatures when embryos could freeze (chicken embryos freeze at −2°C; Lundy 1969). Even during the warmest part of the day, the proportion of time spent incubating was generally greater than 50%. Full-day nest attentiveness ranged from 76% in 2010 and 2011 to 80% in 2005. This is comparable with or higher than species in other warmer systems; Conway and Martin (2000b) estimated 75% mean daily attentiveness for 95 passerine species in North America.
Camfield et al. (2010) indicated that Horned Larks are well suited to breeding in challenging alpine habitats and generally appear able to maintain high levels of attentiveness, even under harsh conditions. The additional years of data presented in this study suggest that in this unpredictable environment larks are faced with conditions in some years that push the limits of their coping abilities. Obtaining statistical confidence and interpreting patterns during extreme years that occur occasionally is problematic. We should avoid placing too much emphasis on the significance of the strong declines in embryo hatchability and overwinter survival of Horned Larks following a year of cold temperatures. However, it is important to recognize that extreme environmental conditions are likely to increase in frequency and severity in high-elevation habitats (IPCC 2013). As a result, alpine-breeding birds may soon be confronted regularly with environmental conditions that could impose considerable fitness costs.
Acknowledgments
We thank C. Ames, A. Clason, M. Grabowski, M. Martin, M. Mossop, L. Sampson, C. Storey, M. Tomlinson, L. Wenn, and M. Wong for field assistance. We also thank C. Cooper and H. Mills for providing program Rhythm 1.0 and W. McKenzie for providing critical weather data. We are grateful to S. Wilson for conducting survival analyses, A. Norris for providing valuable statistical advice, and S. Hinch, D. Weary, and two reviewers for providing helpful comments on earlier manuscript drafts. This work was supported by the Natural Sciences and Engineering Research Council of Canada (Alexander Graham Bell Canada Graduate Scholarship to E. C. MacDonald, Postgraduate Doctoral Scholarship to A. F. Camfield, and Discovery and Northern Research Supplement research grants to K. Martin), the Northern Scientific Training Program, the American Ornithologists' Union (Student Research Award to A. F. Camfield), Environment Canada (Science Horizons internship to A. F. Camfield), and the University of British Columbia (graduate fellowships to E. C. MacDonald and A. F. Camfield). All research protocols were approved by the University of British Columbia Animal Care Committee (AUP numbers A07-0048 and A10-0128).
Literature Cited
