Sources of variation in winter basal metabolic rate in the great tit


†Author to whom correspondence should be addressed. E-mail:


  • 1Basal metabolic rate (BMR) is the most widely used standard measurement of the cost of living. Despite the acknowledged phenotypic flexibility of BMR, little is known about the patterns of variation in wild animal populations.
  • 2We studied the sources of variation in BMR of great tit Parus major (L.) among individuals from two wild populations: Oulu (northern Finland) and Lund (southern Sweden) during six consecutive years.
  • 3By means of a multivariate approach, we found year, locality, date, previous week average minimum temperature, age, body mass, and the interaction between locality and year were the factors retained in the final model, together explaining 71·1% of the total variation in BMR.
  • 4Birds from Oulu (n = 168) had a higher BMR than Lund birds (n = 156), and their BMR varied more between years than that of Lund birds. The two populations reacted in the same way to the other sources of variation examined.
  • 5Great tits from both populations showed a positive relationship between BMR and body mass and a negative relationship between BMR and date, previous week average minimum temperature and age.
  • 6This study highlights the need to standardize BMR measurements when testing predictions about metabolic rates from individuals of wild populations.


Despite the large amount of published data on the basal metabolic rate (BMR) of various species, information about the factors, apart from body mass, that cause variation in BMR in wild populations is scarce. The few studies that have been performed are mostly restricted to interspecific analyses (McNab 1988; Rezende, Bozinovic & Garland 2004), with intraspecific analyses mostly confined to laboratory-reared animals (Ksiazek, Konarzewski & Lapo 2004; Labocha et al. 2004; Sadowska et al. 2005). Recent evidence from some of these studies suggests that intraspecific variation in BMR has a strong genetic component (Wikelski et al. 2003; Ksiazek et al. 2004; Sadowska et al. 2005). However, BMR also appears to be an extremely plastic character, which varies not only between populations but also at an individual level (Bech, Langseth & Gabrielsen 1999; Kerimov & Ivankina 1999; Horak et al. 2002; Labocha et al. 2004; Speakman, Król & Johnson 2004). In addition to genetic and environmental effects, intraspecific variation in BMR may be due to developmental factors or to strategic decisions (Dawson et al. 1983; Burness, Ydenberg & Hochachka 1998; Swanson & Olmstead 1999; Nilsson 2002; Bech et al. 2002; Wikelski et al. 2003; Broggi et al. 2005).

Individuals of endotherm species living year-round at temperate to arctic latitudes are subject to special environmental factors. In order to survive, such individuals have to cope with marked seasonal changes in both food availability and energy requirements for thermoregulation. In winter, both non-renewable food resources and the time available to obtain them decrease, while thermostatic costs increase. Such factors combine to make winter an energetically stressful period, especially for day-feeding, arboreal species such as small birds, which undergo a process of winter acclimatization consisting primarily of a metabolic improvement in thermogenic capacity and endurance (Dawson & O’Connor 1996; Pravosudov & Grubb 1997). However, to fulfil such metabolic requirements birds need to acquire and process more food, normally at the expense of an increased energetic cost of living (Lindström & Kvist 1995; Dutenhoffer & Swanson 1996).

Many studies have shown that winter conditions play a proximate role in regulating metabolism. In free-living as well as in laboratory-acclimated small birds, mass-specific BMR is normally higher in winter than in summer (for review see Swanson in press), and widely distributed species often show a negative relationship between BMR and winter harshness (Kendeigh & Blem 1974; Broggi et al. 2004). It has also been suggested that constraints on the energy expenditure that is necessary to compensate for winter conditions ultimately determine species distribution borders (Root 1988; but cf. Dawson et al. 1983; Repasky 1991; Canterbury 2002).

Understanding how energy is managed among individuals, and its relevance in shaping life histories and distribution ranges, requires knowledge of the sources of variation in BMR in wild populations. Further, the study of environmental effects on the expression of, and selection pressures on, ecophysiological characters in wild populations is of crucial importance in understanding patterns of colonization into new areas and reaction to changes in climate (Hoffmann & Blows 1994).

In previous studies we have shown that great tits, Parus major (L.), from two populations with contrasting winter conditions, are locally adapted to respond metabolically to the environmental conditions (Broggi et al. 2004; Broggi et al. 2005). However, the environmental factors responsible for these patterns of variation between populations remain unresolved. The aim of this study was to determine sources of among-individual variation in BMR during the non-breeding season in wild populations of great tits. Such analyses have not, to our knowledge, been performed for any wintering bird population, but are important for isolating the background variables that define the null hypothesis against which other factor(s) of interest can be tested. We further analysed whether birds from the two populations are affected differently by such variation in environmental conditions.


To study the sources of interindividual variation, we analysed BMR measurements of individual birds captured in the vicinity of Oulu, Finland (n = 168 individuals) and Lund, Sweden (n = 156 individuals) from winter 1999–2000 until winter 2004–05 (last winter data available only from Oulu). Birds were sexed and aged (as yearling or older) following Jenni & Winkel (1994) and otherwise processed by standard methods (Broggi et al. 2004). Birds handled in their second or later winter were aged by means of recapture information, and their age was then categorized in winters (from October to March) (first winter = 1, first winter or older = 2, second winter = 3, etc.). Only one measurement per individual was used in the analyses: when birds were recaptured in later years, only the last measurement was chosen in order to balance sample size in age classes. Similarly, when one individual was measured more than once within a year, a single measurement was chosen to balance the sampling throughout the year.

Winter conditions near Lund (55°40′ N, 13°25′ E) are milder than in Oulu (65° N, 25°30′ E), minimum temperatures on capture days ranged from −8 to +6 °C, vs−24 to +7 °C, and average minimum temperatures between November and March also differed (Table 1). The difference in latitude between the two study areas is accompanied by differences in day length during mid-winter; 7 h in Lund and 3·5 h in Oulu. Great tits in the Lund study area live year-round in mixed deciduous forests fragmented by agricultural landscapes, and do not rely on feeders for survival. In contrast, great tits in Oulu breed in mixed deciduous–coniferous forest, and overwinter close to human settlements that provide food on which they are probably highly dependent for survival (Orell 1989). More details on the study areas and trapping procedures are provided elsewhere (Broggi et al. 2004).

Table 1.  Average minimum temperature (°C) from 1 November to 1 April in Oulu and Lund during the study period
  1. No capture data available during winter 2004/05 in Lund.


metabolic measurements

The BMR is defined as the average minimal oxygen consumption under post-absorptive digestive conditions during the resting phase of the daily cycle of non-growing, non-reproductive animals at thermoneutrality (McNab 1997). Thus BMR was measured in terms of oxygen consumption during the night in an open-circuit respirometer in a dark climate cabinet at a constant temperature of 25 °C. The thermoneutral zone for winter-acclimatized great tits has previously been shown to expand over a 20° range (+15 to +35 °C), in line with previous studies on other smaller species and those from lower latitudes (Dawson & Carey 1976; Reinertsen & Haftorn 1986). Both respirometers and data-extraction procedures have been described in detail in previous studies (Broggi et al. 2004; Broggi et al. 2005). After one night of measurements, birds were released at the point of capture. All procedures were conducted in agreement with local ethical committees.

statistical analyses

We used a general linear model to explain the variation in BMR with locality, year and sex as categorical predictors. Date (1 October = 1), day length (min), body mass (g), wing length (mm), tarsus length (mm), age (1 = first winter), previous week's average minimum temperature (°C) and average minimum temperature (°C) for the period 2–4 weeks before capture were included in the analyses as covariates. As we were specifically interested in variation between the sites, we included first-order interactions between locality and the other variables in the full model. Date squared was also included as a predictor variable in order to test for non-linear effects of date on BMR. The full model was reduced in a stepwise manner according to the highest P value starting with the interactions, and Akaike's information criterion (AIC) was used to choose when to stop the stepwise elimination process, thus ending up with the final model (Quinn & Keough 2002). Sequential reintroduction of each eliminated main effect back into the final model never improved the fit (as determined by AIC). All P values are two-tailed. All continuous variables fulfilled the requirements of normality.


Body mass was the most important predictor of overall BMR, heavier birds having higher rates of basal metabolism (Table 2). Birds from Oulu were significantly heavier [Oulu mean ± SD (N): 19·48 ± 1·59 (168); Lund: 18·34 ± 0·89 (156); t322 = 7·85, P < 0·001] and had longer tarsi [Oulu: 19·45 ± 0·86 (166); Lund: 19·01 ± 0·85 (129); t293 = 4·40, P < 0·001] than birds from Lund, but individuals from the two populations did not differ in wing length [Oulu: 75·63 ± 2·81 (166); Lund: 75·7 ± 2·24 (129); t293 = −0·24, P = 0·81].

Table 2.  The final model (as judged from the lowest AIC) of a general linear model analysis on basal metabolic rate for great tits measured in Oulu and Lund.
PredictorsB ± SE% Var.FP
  1. The following predictors were categorical: locality, winter and sex. The following predictors were continuous: date (1 October = 1), squared date, day length (min), previous week average minimum temperature (°C), average minimum temperature (°C) for the period 2–4 weeks before capture, winter (November–March) average temperature (°C), body mass (g), wing length (mm), tarsus length (mm), age (1 = first winter) and first-order interactions between locality and the other variables. Partial regression coefficients (B) with corresponding SE and partial F and P values are given for each independent variable. Proportion of variation explained (% Var.) is the amount by which overall r2 increases when the respective variable is included in the regression model. The final model explained 71·1% of the variation on BMR (N = 324).

Body mass 0·046 ± 0·00540·692·14<0·001
Date−0·001 ± <0·001 1·637·72<0·001
Average week min. temp.−0·006 ± 0·001 2·719·71<0·001
Age−0·011 ± 0·003 1·614·78<0·001
Locality 16·615·77<0·001
Winter  5·113·28<0·001
Winter × locality  2·9 7·75<0·001

Birds from Oulu had a higher overall BMR than birds from Lund across the winter and throughout the years (Table 2; Figs 1 and 2), locality being the second-best predictor. Furthermore, differences between years explained a significant portion of the observed variation in BMR (Table 2; Fig. 2). To see if this was mainly an effect of average minimum winter temperatures (Table 1), we added these to the final model. However, the winter temperature did not explain any significant part of the variation in BMR (F1,308 = 0·14; P = 0·7). We also added average minimum winter temperature, instead of year, to the final model. The winter temperature became significant, but the AIC increased substantially, showing that including year per se resulted in a better model than including winter minimum temperature. Birds from both populations increased their BMR as temperature during the preceding week decreased (Table 2; Fig. 3) and as winter progressed (Table 2; Fig. 1). The reason for date being a better predictor than date squared is probably due to few being measured before the winter solstice. Furthermore, we found a significant interaction between locality and year (Table 2; Fig. 2), which appears to be dependent on larger variation in mean BMR between years in Oulu than in Lund. Finally, older birds had a lower BMR (Table 2; Fig. 4). The inclusion of other interactions with locality from the final model indicated that BMR of great tits in the two populations reacted in the same general way to the environmental and intrinsic variables that we used as independent factors.

Figure 1.

Relation between individual basal metabolic rate and date (1 = 1 October) among great tits from the populations in Oulu (•) and Lund (○). Second-order regression lines are plotted for each locality.

Figure 2.

Mean basal metabolic rate (+SE) for Oulu birds (filled bars) and Lund birds (open bars) during six consecutive winters.

Figure 3.

Relationship between individual basal metabolic rate and previous week average minimum temperature (°C) among great tits from the populations in Oulu (•) and Lund (○). Linear regression lines are plotted for each locality.

Figure 4.

Mean basal metabolic rate (BMR) residuals (±SD) for Oulu birds (•) and Lund birds (○) of different age classes. Residuals were obtained for each population after regressing all significant predictors except age on BMR values. Age class 1 = first winter; 2 = first winter or older; 3 = second winter; 4 = second winter or older, etc.).


In line with a previous study (Broggi et al. 2004), great tits from the more northerly population had a higher BMR than individuals from the southern population, and this difference persisted after accounting for several other significant variables affecting variation in BMR. Despite previous evidence showing that BMR differences between populations appear to be intrinsically determined (Broggi et al. 2005), individuals from the two populations appear to be similarly affected by the set of variables that we tested, suggesting that both populations have similar patterns of variation in relation to these factors, albeit at different levels of basal expenditure. The different levels of expenditure are probably dependent on the rich and stable food resource provided by humans in Oulu (Broggi et al. 2003). However, some of the differences in BMR between Oulu and Lund also appear to be embedded in the factor year. We can only speculate about what characteristics of winter are important for BMR. Although average annual minimum temperature probably plays an important part, year was a better predictor than winter temperature, indicating that other characteristics, such as precipitation, air humidity, food availability or predation risk, may be involved. The larger variation in BMR between years in Oulu compared with Lund may indicate a larger variation in the harshness of winter conditions at higher latitudes (for a discussion of such variation see Broggi et al. 2003). Variation between (and maybe also within) years may lead to selection for different reaction norms in the two populations, as found by Broggi et al. (2005).

Alternatively, between-year variations in BMR could be due to factors occurring prior to the winter itself, but with an expression of the effects being delayed. For example, variation in an individual's condition after the previous breeding season may affect plumage quality (Nilsson & Svensson 1996), or impose physiological stress that could, in turn, affect the subsequent winter energy budget (Dawson & O’Connor 1996; Beckman & Ames 1998).

Among the size variables, body mass explained the largest portion of variation in BMR, as previously found by Hulbert & Else (2000). Thus mass is probably the best predictor of the size of metabolically active tissues such as the alimentary tract, liver and kidneys. (Alexander 1999). Furthermore, energy management in small passerines has traditionally been studied from the point of view of the acquisition and maintenance of energy reserves (Rogers & Smith 1993; Grubb. & Pravosudov 1994; Biebach 1996), and ‘winter fattening’ has been suggested as one of the main strategies to counteract the increased energy demands during winter (Witter & Cuthill 1993; Pravosudov & Grubb 1997; but cf. Broggi et al. 2003). Indeed, correlated increases in BMR and body mass could be part of the same response to winter conditions, that is, winter acclimatization (Dawson & O’Connor 1996).

Two environmental variables (temperature and date) were included in the final model, together explaining about 5% of the overall variation, and could be regarded as representing short-term and long-term conditions, respectively. Short-term environmental conditions appear to be more important in explaining BMR variation than conditions experienced some time ago, probably reflecting the plastic nature of the BMR as a character (Swanson & Olmstead 1999). However, predictable changes in environmental conditions with winter progression, such as depletion of natural food resources, also appear to play a role in explaining the observed variation.

Sex did not explain any significant part of the variation in BMR, in line with other studies of wintering birds (Wikelski et al. 2003; Broggi et al. 2004). However, we found a significant decline in BMR with age in both populations, in agreement with Kerimov & Ivankina (1999). Two non-mutually exclusive mechanisms could explain this result. First, as previously found in humans and laboratory mice, BMR within individuals could decline with age (for review see Hulbert & Else 2000). Second, individuals with a consistently low BMR throughout their life might possibly experience an increased life span, leading to a higher prevalence of ‘low-BMR’ individuals at older ages. Both alternatives would be in agreement with theories of ageing invoking a trade-off between energy metabolism and life span. This would result from either a trade-off between energy expenditure and energetically expensive processes for self-repair mechanisms, or a direct trade-off between metabolism-driven production of detrimental free oxygen radicals and life span (Beckman & Ames 1998). Thus our results suggest that birds with high BMR have a selective advantage in cold environments, but suffer from an accelerated ageing process, potentially leading to different optimal life-history strategies depending on the winter environmental conditions.

This study highlights the need to standardize BMR measurements when comparing either populations or individuals within a population. When testing a specific hypothesis that includes measurements of BMR, it is important to take into account other sources of variation in BMR.


We are indebted to Kent Andersson, Paavo Bergman, Anna Gamero, Sara Henningsson, Jussi Kupari, Tanja Nikowitz and Kristjan Nitepold for their help in the field. We also want to thank J. Korva for supplying some of the environmental data. Petri Kärkkäinen, Matti Rauman and the staff from the Oulu University Research Facility provided technical support and field assistance, and we are grateful for that. We also wish to thank Claus Bech for enriching discussions and help. Supported by the Academy of Finland project no. 102286 and 47195, and the Thule Institute of the University of Oulu.