Female zebra finches compromise clutch temperature in energetically demanding incubation conditions


Correspondence author. E-mail: andreas.nord@zooekol.lu.se


1. Avian embryos depend on the incubating parent to provide a thermal environment suitable for embryogenesis, but as the maintenance of optimal incubation temperatures is energetically costly, an incubating bird often must trade off embryonic investment against self-maintenance.

2. We manipulated the energetic cost of incubation in female zebra finches (Taeniopygia guttata Vieillot) by varying ambient temperature and clutch size during nocturnal incubation and recorded the corresponding effects on incubation metabolic rate and clutch temperature.

3. Females increased their night-time incubation metabolic rate more than twofold when incubating at 10 °C compared to when incubating close to thermoneutrality (28 °C). Furthermore, clutch enlargement caused females to elevate their metabolic rate with 2·8% per additional egg added to the clutch.

4. However, despite spending more energy, females did not fully cover the increased costs of incubation, because clutch temperature decreased with decreasing ambient temperature and increasing clutch size.

5. These findings suggest that parental investment in incubation can be energetically constrained and sometimes result in clutch temperatures below the optimal level for embryonic development, at least during nocturnal incubation.


The incubation period for most birds is characterized by comparatively low levels of parental activity and reduced thermoregulatory costs while sitting on the eggs in well insulated nests. Nonetheless, an incubating bird must invest considerable amounts of energy to keep eggs at sufficient temperatures and so normally experiences substantially higher energy expenditures than non-incubating congeners (Williams 1987, 1996; Williams & Dwinnel 1990; Tatner & Bryant 1993; Tinbergen & Williams 2002; but see Engstrand, Ward & Bryant 2002 for evidence of no such effects). In some cases, the energetic cost of incubation even approaches that for birds feeding young (Moreno & Sanz 1994; Williams 1996, 2001).

The energy required to maintain eggs at suitable temperatures varies depending on the physical attributes of the incubation environment (Williams 1996; Thomson, Monaghan & Furness 1998). Thus, parents normally experience increased metabolic demands when incubating enlarged clutches or in colder environments (Biebach 1979, 1981, 1984; Vleck 1981; Haftorn & Reinertsen 1985; Weathers 1985; de Heij et al. 2007), resulting in an increased daily energy expenditure in strenuous incubation conditions. Similarly, parents allocate less energy to egg heating when incubation costs are experimentally relieved (Bryan & Bryant 1999; Cresswell et al. 2004; Pérez et al. 2008; Ardia et al. 2009; D’Alba, Monaghan & Nager 2009). In line with this, variation in the amount of energy required for incubation has been shown to affect a variety of parental and neonatal fitness components, such as adult survival (Visser & Lessells 2001; de Heij, van den Hout & Tinbergen 2006), future fecundity (Hanssen et al. 2005), chick rearing capacity (Heaney & Monaghan 1996; Reid, Monaghan & Ruxton 2000a,b; de Heij, van den Hout & Tinbergen 2006), hatching success (Moreno et al. 1991; Siikamäki 1995; Reid, Monaghan & Ruxton 2000a,b; Engstrand & Bryant 2002), fledging success (Reid, Monaghan & Ruxton 2000a,b; Ilmonen, Taarna & Hasselquist 2002), fledgling mass (Cichon 2000), nestling growth rate (Pérez et al. 2008; Nilsson, Stjernman & Nilsson 2008) and incubation duration (Coleman & Whittall 1988; Moreno & Carlson 1989; Smith 1989; Siikamäki 1995; Wiebe & Martin 2000; Engstrand, Ward & Bryant 2002; Dobbs, Styrsky & Thompson 2006).

Despite compelling evidence that high incubation demands can affect nestling performance, the mechanisms involved in transferring costs from parents to developing young have received comparatively little attention. Although avian embryos depend on high and stable egg temperatures for normal development (Webb 1987; Nilsson 2006) parents might need to trade-off self-maintenance against providing an optimal environment for embryonic development. If so, it seems likely that variation in egg temperature during incubation can be instrumental in explaining these transgenerational costs of incubation. In line with this, egg temperature seems to be related to natural variation in ambient temperature (Haftorn 1983; Conway & Martin 2000) and recent experimental data suggest that clutch temperature may vary predictably with the energetic cost of incubation (Niizuma et al. 2005; Ardia et al. 2009).

The inference of variation in nestling quality in response to incubation demands is often confounded by the fact that treatment effects may simultaneously operate both during the pre- (i.e. embryonic) and/or post-hatching (i.e. brood rearing) period, but few studies have attempted to isolate effects of incubation conditions from effects on parental behaviours after hatching (but see de Heij, van den Hout & Tinbergen 2006; Nilsson, Stjernman & Nilsson 2008; Pérez et al. 2008). Thus, it is currently uncertain whether observed fitness decrements in nestlings is best explained by a sub-optimal nest microclimate during incubation or by reduced parental effort during brood rearing, or by a combination of the two. To distinguish between these effects, it is therefore necessary to quantify variation in embryonic temperature in conjunction with parental energetic demands during incubation and to separate potential carry-over effects on nestling phenotype from parental behaviours after hatching. In addition, measurements of clutch temperature may more adequately reflect parental investment in incubation (at least when comparing clutches under identical environmental conditions), because it directly reflects energy allocation to the developing embryos.

In this study we simultaneously measured metabolic rate and clutch temperature for female zebra finches (Taeniopygia guttata Vieillot; Fig. 1) incubating in different ambient temperatures and with different clutch sizes. To account for potential between-female differences in incubation effort, we exposed all females to all experimental treatments in a random order and subsequently analysed data in the context of individual variation. Our overall aim was to assess the relationship between incubation metabolic rate and clutch temperature over a range of thermal environments. Specifically, we were interested to see if this relationship was negative and thus could help to explain the often observed relationship between the energetic cost of incubation and nestling performance (see above). We hypothesized that incubation metabolic rate would increase with decreasing ambient temperature and with increasing clutch size, but that clutch temperature would be inversely related to incubation demands. This is, to our best knowledge, the first study to provide simultaneous records of metabolic rate and clutch temperature during incubation in a passerine bird.

Figure 1.

 Female zebra finches Taeniopygia guttata Vieillot. Photograph taken by Sandra Chiriac.

Materials and methods

Animal husbandry

A total of 23 zebra finch pairs with previous breeding experience were used in the study. About half of the birds (n = 13 pairs) were already established pairs that had bred together before, whereas the remainder were brought together for the first time at the start of the experiment. Of these, 13 pairs (12 established, 1 novel) built a nest, laid eggs and commenced incubation within the experimental period. The female in the newly formed pair produced a smaller clutch (two eggs) than did females in established pairs (mean = 4·0 ± 0·25 eggs). However, results were not influenced by this variation, because regression coefficients and R2-values were not affected when data for the female from the novel pair were excluded from analyses. The pairs were housed separately in cages measuring 60 × 45 × 40 cm under an artificial 13 : 11 light : dark regime at a temperature of approximately 20 °C throughout the experiment.

Birds were provided with mixed seeds, water and cuttlebone ad libitum. When the experiment started, birds were supplied with a nest box (7 × 7 × 12 cm) and nesting material (dried grass and cotton wool). To ensure that potential energy limitations did not affect optimal incubation strategies, the original diet was enhanced with homogenized hard boiled hens’ eggs ad libitum from the second morning following clutch completion and onwards.

Clutch and nest box temperature

On the morning of the fourth day following the start of incubation, clutches were removed and replaced with artificial eggs. Four eggs, which is the modal clutch size in our population (range: 1–6 eggs), were added to each nest, irrespective of the number of eggs laid. When replacing the clutch, small temperature data loggers (I-Button DS1922, Ø 16·5 mm, accuracy = 0·5 °C, Maxim Integrated Products Inc., Sunnyvale, CA, USA), measuring temperature in 1-min intervals, were placed beneath the eggs to monitor clutch temperature. Identical temperature loggers were attached to the roof of all but four of the nest boxes to monitor nest box temperature. All pairs seemed insensitive to this manipulation, and resumed incubation within minutes of having their clutch replaced.

Incubation effort

Incubation effort was determined by measuring night-time incubation metabolic rate in an open circuit respirometer. Four parallel channels with identical setups were used, thus allowing four birds to be measured simultaneously.

One hour after lights being switched off, the nest box entrance was blocked to trap the female inside the nest. The nest box with the female inside was then put into a sealed 1·6 L metabolic chamber and placed in a temperature controlled climate cabinet (Heraeus Vötsch BK600; Vötsch Industrietechnik, Balingen, Baden-Württemberg, Germany) for metabolic measurements. A measurement session ended 1 h prior to lights being switched on (giving a measurement period of 9 h), at which point the nest boxes containing the females were transferred back to their original cages. An incubating female was measured in three different incubation regimes with varying energetic demands; (i) close to thermoneutrality (28 °C; Calder 1964; Cade, Tobin & Gold 1965) with four eggs, (ii) at 10 °C with four eggs and (iii) at 10 °C with six eggs. Clutches were enlarged with two eggs at noon on the day preceding measurements with six eggs. The control clutch size (i.e. four eggs) was restored 24 h later. All measurements were performed in a randomized order and all measurement series were commenced between days 5 to 7 in the incubation period. Females were kept in the same respirometer chamber during all measurements. Neither males nor females appeared to respond negatively to experimental treatment and incubation proceeded normally throughout the experimental period.

Oxygen concentration of effluent sample air scrubbed on silica gel was analysed by a Servomex 4100 oxygen analyzer (Servomex Inc., Crowborough, England) and was automatically registered on a data logger (Grant Squirrel 1202; Grant Instruments Ltd., Cambridge, England) every minute throughout a measurement session. The oxygen analyzer was calibrated against outside air to 20·95% O2 prior to all measurement sessions. A flow rate of 12 L h−1 was maintained using a Bronkhorst HI-TEC mass flow meter (Bronkhorst High-Tech B.V., Ruurlo, the Netherlands). Each measurement run lasted for 20 min, and manual checking for drift confirmed that oxygen consumption was stable throughout the full length of the run in all cases. Oxygen consumption (mL O2 min−1) was defined as the difference in oxygen concentration between effluent air from the metabolic chambers and reference air from the outside according to equation C in Hill (1972). The value of oxygen consumption used in the analyses was taken as the single lowest value from running 10 min averages excluding the periods 2 h subsequent to the start and prior to the end of a measurement session. Data on oxygen consumption was converted to metabolic rate (kJ day−1) by assuming an energy equivalence of 20 J (mL O2)−1.

Birds were weighed to the closest 0·1 g on the day of being supplied with a nest and on the first and last days in the measurement series, using a Pesola spring scale (Pesola AG, Baar, Switzerland). Weighing occurred at the same time of the day at each occasion. In all analyses, mass was considered to be the average of the mass at the first and last day of incubation metabolic rate measurements.

Statistical analyses

Because of malfunction of one of the respirometer channels, data for one female was excluded from analyses. Moreover, three recordings of clutch temperature from the 10 °C/4 egg treatment were not included in the analyses, because females had removed the temperature logger during the respirometer session. In order to avoid variation due to starting and ending of a measurement session, temperature data from the 2 h subsequent to the start and prior to the end of a measurement session were not included in the analyses.

All statistical tests were performed using R 2.6.2 for Windows. Data were graphically checked for parametric assumptions and the presence of outliers and high leverage points prior to all analyses. In addition, Levene’s test of homogeneity of variances and the Weisberg-Bingham test of normality was applied whenever applicable. Neither tests nor graphical analyses indicated that any observation violated requirements for linear analysis.

We analysed variation in incubation metabolic rate and clutch temperature, respectively, with linear mixed effects models fitted with restricted maximum likelihood methods (using the lme function in the nlme package), with incubation regime as a covariate and female intercept as a random effect. Female mass was included as a covariate in all analyses of incubation metabolic rate, but it did not explained any variation in the dependent variable (P > 0·2 in all cases) and was accordingly excluded from the final models. To account for possible variation in individual responses to the experimental treatment, we subsequently fitted additional mixed effects models with a diagonal variance–covariance structure with both slope and intercept as random effects. Random slope and random intercept models were compared using likelihood ratio tests and AIC values (Pinheiro & Bates 2004). In all cases, regression diagnostics indicated that the simpler random intercept models should be preferred (P > 0·4 and ΔAIC > 2·0 in all cases).

All means are presented with their standard errors (mean ± SEM) and all significances are two-tailed.


Clutch and nest box temperature

Average clutch temperature varied significantly with ambient temperature (F1,20 = 66·7, P < 0·001, Fig. 2) when females incubated four egg clutches. Females maintained significantly higher clutch temperatures when incubating at 28 °C (mean = 34·7 ± 0·45 °C) than when incubating in room temperature (∼20 °C; mean = 32·8 ± 0·61 °C) and at 10 °C (mean = 30·4 ± 0·89 °C), and when incubating in room temperature compared to when incubating at 10 °C (Tukey’s HSD test: P < 0·05 in all cases). Clutch temperature was significantly higher than nest box temperature during all measurements [mean increase; 28 °C: 5·8 ± 0·49 °C; room temperature (∼20 °C): 10·6 ± 0·77 °C; 10 °C: 16·5 ± 1·21 °C; paired t-test; 28 °C: t8 = 11·9, P < 0·001, room temperature: t7 = 13·8, P < 0·001, 10 °C: t5 = 13·6, P < 0·001].

Figure 2.

 Clutch temperature (°C) during night-time incubation in four different incubation regimes in zebra finches.‘∼20 °C/4 eggs’ refers to incubation in room temperature prior to all experimental treatments. Lines represent single individuals. Bars show means. Note that values for clutch temperature in the 10 °C/4 eggs treatment are lacking for three individuals (as described in the Materials and methods section).

Clutch temperature also varied with clutch size, with females maintaining the clutch at significantly lower temperatures when incubating enlarged clutches (i.e. six eggs; mean = 28·4 ± 0·90 °C; F1,8 = 16·9, P = 0·0034, Fig. 2). As before, clutch temperature when incubating 6 eggs at 10 °C was significantly higher than nest box temperature (mean increase = 13·8 ± 1·23 °C; paired t-test; t8 = 11·2, P < 0·001).

Incubation effort

Incubation metabolic rate was significantly lower when females incubated a clutch of 4 eggs at 28 °C (mean = 20·47 ± 0·50 kJ day−1) than when they incubated at 10 °C with the same clutch size (mean = 43·27 ± 1·04 kJ day−1; F1,11 = 402·1, P < 0·001, Fig. 3). In addition, females spent more energy when incubating six eggs at 10 °C (mean = 45·67 ± 0·69 kJ day−1) compared to when incubating four eggs at the same temperature (F1,11 = 13·6, P = 0·0036, Fig. 3). This difference amounts to approximately a 2·8% increase in female energy expenditure per additional egg.

Figure 3.

 Night-time incubation metabolic rate (kJ day−1) during three different incubation regimes in zebra finches. Each line corresponds to a single individual. Bars show means.


Our results show that the energetic cost of incubation increased with decreasing ambient temperatures and increasing clutch size, suggesting that female incubation effort was not constant but increased with increasing demands. However, females only partially met the increased energetic requirements, because clutch temperature decreased with decreasing ambient temperature and increasing clutch size (Fig. 2). It is therefore possible that females had to trade off their own effort against maintenance of clutch temperature. This corroborates recent experimental findings showing that incubating females moderate their investment in egg heating in relation to the assumed level of energetic stress during incubation (Ardia & Clotfelter 2007; Ardia et al. 2009). Thus, the allocation of effort to incubation seems to constrain the possibility to meet the energetic costs of keeping eggs warm, and the parental care provided by females while incubating may sometimes be below the optimal level for embryogenesis.

Zebra finches incubating in demanding environments will thus be directly affected by factors related to increased energy expenditure, such as increased risks of oxidative stress (Alonso-Alvarez et al. 2004; Monaghan, Metcalfe & Torres 2009) and indirectly affected by reduced offspring quality due to temperature dependent constraints on embryonic development (Olson, Vleck & Vleck 2006; Hepp, Kennamer & Johnson 2006; Nilsson, Stjernman & Nilsson 2008). Thus, an elevation of incubation demands could potentially affect fitness components of both mothers and offspring. As zebra finch females incubate unassisted during the night (but not during the day; Zann & Rossetto 1991), they need to maintain nocturnal energy expenditure at a level compatible with their endogenous energy reserves, which could potentially compromise the maintenance of clutch temperature. Thus, patterns of clutch temperature regulation and incubation metabolic rate are not necessarily comparable between nocturnal and diurnal incubation or between species with single-sex and bisexual incubation (cf. Nilsson 1993), provided that males relieve females from incubation duties when her reserves reach some threshold level (Cresswell et al. 2003).

Females increased metabolic rate with 2·8% per additional egg when incubating enlarged clutches, which is in close correspondence with results from previous studies on zebra finches (Vleck 1981) as well as on other passerines (Biebach 1981, 1984; Haftorn & Reinertsen 1985; Weathers 1985; de Heij et al. 2007). This increase in energy expenditure is rather small in absolute terms and it seems unlikely that this cost alone could restrain a female from laying an additional egg and thus serve as an important agent in the evolution of avian clutch size (cf. Monaghan & Nager 1997). In line with this, clutch enlargement does not always elevate overall daily energy expenditure during incubation (Moreno & Carlson 1989; Engstrand, Ward & Bryant 2002; de Heij et al. 2008), but may nonetheless elevate the instantaneous metabolic cost of keeping eggs warm (de Heij et al. 2007) and decrease parental fitness (de Heij, van den Hout & Tinbergen 2006). In addition, results from this work suggests that females do not always pay the full energetic cost of incubating additional eggs, because maintaining equilibrium egg temperatures also when incubating enlarged clutches would presumably require a substantial increase of parental energetic investment. Thus, the true energetic cost of incubating larger clutches might therefore have been underestimated both here and in previous work, as females might not have been prepared to increase clutch temperature to levels experienced by smaller clutches. Consequently, it is possible that females are energetically constrained from incubating large clutches, despite the normally relatively modest increases in energy turnover rates.

However, clutch size manipulations sometimes produce undesired experimental artefacts, because females might be physically unable to efficiently cover all eggs in an enlarged clutch (Reid, Monaghan & Nager 2002). This can result in increased intra-clutch temperature variability (Reid, Monaghan & Ruxton 2000b) and reduced overall clutch temperatures (Niizuma et al. 2005). However, it is unlikely that the inverse relationship between clutch size and clutch temperature as found in this study was a result of physical constraints as the temperature logger was placed directly beneath the brood patch during all measurements and thus should directly reflect the amount of energy allocated to egg warming. Moreover, enlarged clutches were within the natural range of variation (see above), which further suggest that females were physically able to provide ample heat also to these clutches.

Ambient temperature strongly affected the energetic cost of incubation, with metabolic rate increasing by 111% during the transition from 28 °C to 10 °C. This was not merely a result of the higher thermoregulatory costs, because the fact that females altered patterns of energy allocation away from egg heating in low ambient temperatures indicates that at least some of the higher metabolic rate in the 10 °C treatment truly was caused by elevated costs of incubation. The importance of ambient temperature in shaping incubation investment is further illustrated by a number of experimental studies wherein manipulations of the thermal environment experienced by incubating parents have been shown to affect parental investment in incubation with regards to both patterns of nest attentiveness and adult energy balance (Bryan & Bryant 1999; Reid, Monaghan & Ruxton 2000a,b; Visser & Lessells 2001; Nilsson, Stjernman & Nilsson 2008; Ardia et al. 2009).

We have provided evidence that nocturnal incubation in zebra finches can be energetically costly, resulting in an augmentation of parental investment when incubation demands increase. However, this study also revealed that parents did not fully compensate the increased energetic costs, but instead maintained eggs at lower temperatures in strenuous incubation conditions which might constrain offspring development. As a result, our results might underestimate the true energetic cost of incubation since maintaining high clutch temperatures also in strenuous conditions probably entails a substantially higher energetic cost than found in this study.


We thank Johan Nilsson for kindly providing temperature loggers and Jan Johansson for help with experimental equipment. We also thank Martin Stjernman for statistical advice and Sandra Chiriac for providing zebra finch pictures. Comments from the editor and two anonymous referees improved a previous version of the manuscript. This study was supported by grants from the Swedish Research Council (to J.-Å. N.). All experimental protocols adhere to the guidelines of the Swedish Animal Welfare Agency and were approved by the Malmö/Lund Animal Care Committee.