(a) Capture and handling

This study was carried out within the Forêt d’Enseignement et de Recherche Macpès, Québec, Canada (48° 30’ N, 68° 52 W) between January and March 2010 (n = 56) and from October 2010 to March 2011 (n = 149). Data from summer individuals were collected in August 2010 (n = 12) and 2011 (n = 11). To attract chickadees and facilitate capture, 18 feeding stations were set up within the forest with an average distance between stations of 1.9 km (see 30,32 for a description of the stations). Feeders were regularly filled with black sunflower seeds. On capture days, feeders were removed and homemade potter traps (15 cm x 15 cm x 15 cm) baited with seeds were placed on a tray installed on a wooden fence pole. All birds were caught during morning (between 08:00 and 13:00) and removed from traps within one minute of capture. Weather stations in the forest [35] recorded temperature data allowing us to track ambient temperatures over the two years of the study.

Birds caught for the first time were banded with a USGS numbered metal band as well as a unique combination of three plastic color bands to allow further identification from a distance. For each capture, birds were first weighed then had the length of their beak, head plus beak, tarsus, tail and wing measured [36]. Following these measurements and depending on capture success, up to four birds per day were brought to the field station for metabolic measurement.

(b) Ethics statement

All bird manipulations were approved by the animal care committee of the Université du Québec à Rimouski (CPA-37-09-68) and have been conducted under scientific and banding permits from Environment Canada - Canadian Wildlife Service (Permit Number: 10704H).

(c) Respirometry

Once at the field station, birds were maintained in separate cages (39 cm x 43 cm x 31 cm) supplied with food (sunflower seed) and water ad libitum until measurements were made. Cages were kept in a room receiving natural light through a window and maintained quiet to avoid disturbance. At around 13: 00, we began M_SUM trials by measuring two birds in parallel using Fox, Box oxygen analyzers (Sable Systems, Las Vegas, NV, USA). This was followed by a second trial on the remaining two birds, which began before 15: 00. Briefly, birds were first weighed (± 0.1 g) and body temperature was measured with a thermocouple reader (Omega model HH-25KC, NIST-traceable, Omega, Montréal, Qc, Canada) using a copper-constantan thermocouple inserted into the cloacae approximately 10 mm deep. Then, birds were put in a stainless steel metabolic chamber fitted with a perch and were exposed to helox gas (21% oxygen, 79% helium) using an average flow rate of 1109 ml.min^-1 controlled by mass flow valves (Sierra Instruments, Side-Trak® Model 840 (Monterey, CA, USA)). We recorded oxygen consumption of each bird using a sliding cold exposure protocol [37] with a decrease in ambient temperature of 3°C every 20 minutes, starting at 0°C in winter and at 6°C in summer. We ended the trials when birds became hypothermic, which was easily identifiable in real time as a steady decline in oxygen consumption for several minutes. Body temperature was measured again immediately after taking birds out of their chambers. We assumed a bird had reached its M_SUM when body temperature after a trial was ≤ 38°C [38] (mean body temperature after M_SUM measurement = 33.6 ± 0.2°C). Data from individuals showing a body temperature above this threshold were discarded. Birds were weighed again after measurements and the average body mass was used for the M_SUM analysis. Birds were then brought back to their cage with food and water ad libitum until BMR measurement starting at 19: 00.

Each day, all four birds had their BMR measured simultaneously overnight (from 19: 00 to 06:00). Individuals were maintained at 30°C throughout the trial (within the thermoneutral zone for this species [39]) and received a constant flow of air. Birds were weighed before and after measurements and average mass was used in BMR analyses.

Oxygen analyzers were adjusted each day to 20.95% using CO₂-free dry air and mass flow valves were carefully calibrated for air and helox using a bubble-O-meter (Dublin, OH, USA) once per winter. Metabolic rate calculations were done with ExpeData software, v 1.2.6 (Sable Systems, Las Vegas, NV, USA). M_SUM and BMR calculations were based respectively on the highest and lowest averaged 10 minutes of oxygen consumption per measurement sequence according to Lighton’s equation 10.1 [40]. The instantaneous measurement technique [41] was used for M_SUM while BMR was calculated using the steady state approach. The duration of BMR trials (around 11 hours) insured that birds were post-absorptive at time of BMR measurement (which was obtained after 6 h 40 min ± 8 min of recording on average). Since wintering birds use mostly lipids as substrate for shivering [15,42,43], we estimated energy consumption using a constant equivalent of 19.8 kJ.L^-1 O₂ and converted to Watts [44]. After BMR measurements, birds were put back in their cage with access to food and water until release at capture site around 2 hours later.

(d) Sexing individuals

130 individuals (56 females and 74 males) caught for this study were sexed by PCR [45] or dissection. We determined sex of the remaining 98 birds using their morphometric data in a discriminant analysis [46]. Overall, 85 individuals were identified as females, 107 as males and 36 birds remained undetermined.

(e) Statistical analysis

(a) Temperature

During the first winter of the study (2009-2010), chickadees experienced mean daily temperatures below 0°C from December to March (table 1). The coldest average temperature was recorded in December, and the lowest minimal temperature was measured in February. During the second year (2010-2011), mean temperatures fell below 0°C from November to March and reached their coldest mean ambient and minimal values in January (table 1). However, the following months remained relatively cold as minimal temperatures changed by less than one degree in the next two months. Overall, the second winter of the study was colder than the first and had more months with minimal temperatures below -25°C.

(b) Inter-seasonal variation in body mass and metabolic performance

Peak values in average metabolic performance were recorded in February for both years (see below). We therefore calculated inter-seasonal variation in body mass and metabolic parameters between peak of winter and summer using values measured in February and August.

Average body mass did not vary between years or seasons but positively varied with time of capture (F_1,70 = 11.2, p < 0.01). Males were also on average 10.5% heavier than females (sex: F_2,112 = 59.8, p < 0.0001, males: 12.06 ± 0.07 g; females: 10.91 ± 0.08 g, undetermined 11.31 ± 0.10 g, Tukey: p < 0.0001). BMR, M_SUM and ME were all influenced by body mass (BMR: F_1,112 = 32.5, p < 0.0001; M_SUM: F_1,109 = 29.6, p < 0.0001, ME: F_1,98 = 5.0, p < 0.05). Mass-independent BMR was 5.9% higher in winter relative to summer (season: F_1,116 = 16.5, p < 0.0001, Figure 1A) while average values for mass-independent M_SUM were 34.2% higher at the peak of winter relative to August (season: F_1,115 = 135.5, p < 0.0001). However, this latter effect depended on the year (year*season: F_1,117 = 7.2, p < 0.01). Mass-independent M_SUM was 13.4% higher in the first winter relative to the second, which led to a M_SUM being 41.9% higher than in summer during the first year compared to a 26.4% seasonal difference during the second year (Tukey: p < 0.0001 in all cases) (Figure 1B). Mass-independent ME was 25.0% higher in winter (6.2 x BMR) than in summer (5.0 x BMR, season: F_1,111 = 56.9, p < 0.0001) and this effect was also dependent on the year (year*season: F_1,109 = 5.2, p < 0.05). Mass-independent ME was 11.4% higher in the first winter than in the second with a ME 32.7% higher relative to summer in the first year compared to a 17.4% difference between winter and summer during the second year (Tukey: p < 0.0001 in all cases) (Figure 1C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Inter-seasonal variation of mass-independent BMR, M_SUM and ME.

Data are least square means for BMR (A), M_SUM (B) and ME (C) controlling for year, season, sex and body mass with bird ID as random parameter. Different letters indicate significant difference between seasons.

https://doi.org/10.1371/journal.pone.0068292.g001

(c) Intra-seasonal variation of body mass and metabolic performance

As for the inter-seasonal analyses, we found no year effect on average body mass in wintering black-capped chickadees. However, body mass varied over time within winter (month: F_6,212 = 8.3, p < 0.0001) increasing by 4.0% between October and December (Tukey: p < 0.01) and remained constant until August (Figure 2, Tukey: p = 0.07). Mass also differed between sexes (F_2,218 = 146.8, p < 0.0001) with males being on average 11.5% heavier than females (males: 12.00 ± 0.05 g; females: 10.76 ± 0.06 g, undetermined 11.18 ± 0.07 g, Tukey: p < 0.0001) and, as these birds are fattening up on a daily basis [35,47], body mass was positively affected by time of capture (F_1,225 = 57.1, p < 0.0001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Intra-seasonal changes in body mass.

Data are least square means of body mass controlling for year, month, time of capture, sex with bird ID as random parameter. Different letters represent significant difference between months.

https://doi.org/10.1371/journal.pone.0068292.g002

Mass-independent BMR (body mass: F_{1, 218} = 83.1.2, p < 0.0001) did not change between years but varied within winter (month: F_6,321 = 6.1, p < 0.0001, Figure 3A). BMR progressively increased by 5.9% between October and February (Tukey: p < 0.05) and then decreased by 5.5% between February and March (Tukey: p < 0.0001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Monthly variation at the population and individual levels in mass-independent BMR, M_SUMand ME.

Population data (black line and dots) are least square means for BMR (A), M_SUM (B) and ME (C) controlling for year, month, sex, body mass and bird ID as random parameter. In B and C - dotted line, open dots and uppercase: 1^st year; solid line, black dots and lowercase: 2^nd year. Different letters represent significant difference between months within a year. Predicted values calculated from LMEs (see text for details) are used to visualize data from individuals captured more than once (grey lines).

https://doi.org/10.1371/journal.pone.0068292.g003

As for BMR, mass-independent M_SUM (body mass: F_1,204 = 97.6, p < 0.0001) did not differ between years but varied within winter (month: F_6,270 = 37.1, p < 0.0001). Average M_SUM increased by 25.0% between October and February and weakly decreased by 5.2% between February and March. However, the month effect also depended on the year (year*month: F_3,266 = 10.0, p < 0.0001). Separated analyses by year showed that the M_SUM peak observed in February was only apparent in 2010, where it reached a value 31.9% higher than our intra-seasonal reference point in October, before declining by 13.3% between February and March (Tukey: p < 0.01). During the second year, mass-independent M_SUM increased steadily throughout winter to reach a peak in March with a value +20.7% higher than that measured in October (Tukey: p < 0.0001) (Figure 3B).

Mass-independent winter metabolic expansibility (body mass: F_1,156 = 8.2, p < 0.01) varied with month (F_6,253 = 18.3, p < 0.0001) with average mass-independent ME increasing by 19.8% between October and March. However, the month effect was also dependent on the year (year*month: F_3,253 = 4.9, p < 0.01). Separated analyses revealed that during the first year, mass-independent ME reached a peak (6.5 x BMR) in February (+27.0% relative to October 2011) before decreasing non-significantly in March (-2.6%, Tukey: p = 0.8). During the second year, mass-independent ME reached its highest value (6.2 x BMR) in March (+19.6% relative to October, Tukey: p < 0.0001) (Figure 3C).

For all metabolic parameters, visual inspection of predicted values for recaptured individuals showed a high level of variability between birds but consistency in their position relative to others (i.e. there were “high” and “low” BMR/M_SUM individuals). Individual patterns were comparable to that observed at the population level (Figure 3).

(d) Relationship between BMR and M_SUM

Linear regression between residuals of whole BMR and whole M_SUM extracted from linear mixed effect models resulted in a significant but weak positive relationship between these parameters (n = 269; r^²_adj = 0.04; p < 0.001). The relationship was not significant when using residuals from models controlling for body mass (p = 0.4).

(a) Inter-seasonal variation in metabolic performance

Mass-independent basal metabolic rate, interpreted here as the energy expenditure of physiological systems remaining active in a resting bird, peaked at 0.27 W in wintering black-capped chickadees. This is comparable to values measured in wintering chickadees from Ohio (0.26 W) and Wisconsin (0.27W) [48] but lower than the ones measured in birds spending their winter in New York (0.29) [49], South Dakota (0.30 W) [48] and Alaska (0.42 W) [50]. It is therefore not surprising to find that the 6% seasonal increase in mass-independent BMR observed here is much lower than the +14% found by Cooper and Swanson in birds from South Dakota [12]. This seasonal variation is, however, in the relatively large range of seasonal changes observed in other temperate free-living resident passerines (from -4.3% in Carpodacus mexicanus [51] to +36.3% in Passer montanus [18]). Assuming the seasonal elevation in BMR reflects an increase in maintenance costs [17,18], it therefore appears that black-capped chickadees from our study site only face a moderate rise in maintenance energy demand in association with seasonal cold acclimatization relative to other populations. If temperature is one of the drivers of winter metabolic performance [10,23], it is likely that differences between populations reflect physiological responses to local conditions.

As for BMR, the seasonal increase in M_SUM (+34%) was in the range of previously reported observations. Seasonal changes in mass-independent M_SUM range from +16.2% in Sitta carolinensis [13,51] to +42.4% in Picoides pubescens [13], with black-capped chickadees from South Dakota showing a 26.3% increase in winter mass-independent M_SUM relative to summer [12].

(b) Intra-seasonal variation in metabolic performance and the uncoupling of BMR and M_SUM

Temperature has been suggested as one of the drivers of winter metabolic performance [10,20,23,48] and this led us to predict a gradual increase in BMR and M_SUM during winter where a peak would be observed during the coldest months of the season. Seasonal variations in BMR were of lower amplitude than expected but changes in average values were consistent with our prediction. Although significant differences were only clear when comparing February with October, November and March, BMR clearly tended to be higher during December, January and February, the coldest months of winter (both years combined, table 1). However, BMR declined rapidly in March, when ambient temperatures were still relatively cold (-16°C and -27°C mean minimum temperature in 2010 and 2011 respectively), which suggests that temperature may not be the sole driver of winter BMR phenotype in chickadees.

In contrast, intra-seasonal changes in average M_SUM differed between years. During the first winter, M_SUM peaked in February and declined in March while in the second year M_SUM increased until the end of our measurements in March. M_SUM variations were therefore consistent with our predictions for the first year, where February was the coldest month based on minimal temperature, but were counterintuitive for the second year where no decline in M_SUM were observed. It must be noted, however, that cold energy-demanding temperatures lasted much longer during the second winter since, although daily temperatures were already warming by March, minimal temperatures were as cold in March as in January. It is therefore likely that birds maintained their thermogenic capacity to its maximal level as long as days with very cold temperature prevailed. As metabolic expansibility is the ratio of M_SUM on BMR, and given the differences in range of intra-seasonal changes in BMR and M_SUM, variations in ME were inevitably affected by changes in M_SUM. It was therefore of no surprise to find similar variation when comparing ME and M_SUM.

The contrast in BMR and M_SUM patterns goes in hand with the hypothesis that these components of metabolic performance respond to different sets of environmental constraints [20,27,28]. Comparing intra-seasonal patterns of change in BMR and M_SUM relative to summer values also suggests a certain level of independence between these parameters. Although it was moderate, the increase in mass-independent BMR between seasons (+5.9% comparing August and February) was the same as that detected between October and February (+5.9%). In fact, the increase in BMR started after November, when birds began to face minimal temperatures below -10°C (table 1), and remained relatively steady until February (Figure 4A). In contrast, the inter-seasonal change in mass-independent M_SUM (+34.2% between August and February) was higher than that measured within season (+25.0% between October and February). In fact, M_SUM had already achieved 21.8% of its inter-seasonal increase when we began our measurements in October (Figure 4B) meaning that, although the largest change in M_SUM appeared between January and February (+45.4% of total inter-seasonal increase, Figure 4B), this parameter began to change well before the beginning of sub-zero mean ambient temperatures (table 1). These results therefore suggest that flexible adjustments in thermogenic capacity appear relatively early in autumn (before October) while physiological components reflected in BMR only begin to change later (from November).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Visual representation of the monthly winter increment in metabolic performance.

Increment in mass-independent BMR (A) and M_SUM (B) relative to summer values. Data measured in August are used as the summer phenotype reference point (August = 0%, black star, dashed line) while maximal levels of BMR and M_SUM recorded in winter are considered representative of the “peak” winter phenotype (February = 100%). Monthly changes in metabolic performance (values above lines) are presented in percent of total inter-seasonal difference. Dotted lines represent BMR variations below the summer reference point.

https://doi.org/10.1371/journal.pone.0068292.g004

Why would BMR begin to increase later in autumn and decline earlier in the spring than M_SUM? The reasons for the uncoupling of these variables are not clear. However, if one considers M_SUM as a variable mainly influenced by the size of shivering muscles [21,31,32,51] and BMR as reflecting changes in size and activity of the main digestive organs in response to cold acclimatization [18,52,53] then it is likely that those metabolic parameters reflect different sets of physiological constraints acting relatively independently, both being associated with winter [54]. Pectoral muscles would start changing early in the season in association with the appearance of cold ambient temperatures (note that sub-zero minimal temperatures were already recorded in October), whereas winter BMR variations would likely reflect changes in diet and in the amounts of food consumed (to sustain an increasing daily energy expenditure, DEE). Chickadees feed mainly on insects during summer but eat up to 50% vegetal matter during winter [55] while food supplementation by feeders may represent only up to 21% of their daily energy intake [56]. Snow typically starts to fall in November at our field site while insects are already visible by March. Therefore, it could be that the combination of a winter increase in DEE and changes in natural diet digestibility and energy content leads to a restructuring of the digestive apparatus, and in turn changes in BMR, appearing independently from those observed in M_SUM. The proximal effect of cold ambient temperature on parameters of metabolic performance [10,23,48] would therefore be much more influential for M_SUM than for BMR. Experimental research is needed to test this “metabolic uncoupling” hypothesis and to determine what biotic and/or abiotic factors trigger seasonal changes in parameters of metabolic performance.

Given the seasonal mismatch in variations of BMR and M_SUM, it is therefore of no surprise to find a lack of significant correlation between these parameters when controlling for body mass in regression analyses. Independence of BMR and M_SUM has also been observed by others [15,27,28,57], but findings remains conflicting [25,33]. Experimental manipulations of BMR and M_SUM, for example by combining diet and temperature treatments, should therefore be conducted to confirm findings.

(c) Intra-individual variation in winter metabolic performance

Studies on seasonal variation of avian metabolic performance are typically conducted at the population level [11,18,30,51] and, although it is rarely stated, they generally assume that population patterns are reflective of those observable within individuals. As far as we know, this is the first study to document with an extensive dataset seasonal variation of metabolism at both the population and individual levels in a resident bird species. Our findings support the common assumption; patterns observed at the population level reflected intra-individual variation in body mass, mass-independent BMR, M_SUM and ME (Figure 3A–C) and are therefore representative of average individual phenotypic flexibility.

(d) Is metabolic expansibility a meaningful variable?

In this study, we considered metabolic expansibility as an indicator of the capacity of an organism to produce heat (because it was based on M_SUM) for a given size of metabolic machinery (which would be reflected by BMR) [12,15,26,28,34]. However, the lack of correlation between BMR and M_SUM, as observed by others [15,27,28,57], as well as the temporal mismatch in these parameters during seasonal acclimatization suggest a lack of a functional relationship between BMR and M_SUM. Although BMR includes the resting energy consumption of physiolgical components involved in active thermogenesis (i.e. muscles), evidences suggest that, in the context of cold acclimatization, its variations are mainly influenced by activity and size of organ systems involved in energy acquisition and digestion [18,30,52], while M_SUM would mostly reflect active energy consumption of shivering muscles [21,31,32]. It therefore becomes apparent to us that metabolic expansibility, as we defined it earlier, may lead to misleading conclusions about machinery adjustments. We therefore suggest caution with the interpretation of this variable and recommend interpreting variations in BMR and M_SUM separately to infer on animals metabolic capacity.