1The aim of this study was to investigate the structure and thermoregulatory function of nests of the Long-Tailed Tit, Aegithalos caudatus.
2The feather lining of Long-Tailed Tit nests represents a major portion (41%) of the total nest mass.
3The mass of feathers varied among nests and declined through the breeding season, but there was no seasonal loss of nest insulation quality because of increasing ambient temperatures.
4In an experiment to investigate the seasonal decline in the feather mass of nests, feathers were added to nests at an early stage of the lining phase of nest construction. Nest structure and insulating properties were then examined following nest completion.
5The total mass of feathers in treatment and control nests did not differ significantly and there was no significant difference in their nest insulation quality.
6Our results demonstrate that Long-Tailed Tits adjust their nest-building behaviour according to the nest's thermal environment. Moreover, nest structure appears to be adjusted to prevailing environmental conditions rather than being a function of feather availability or time constraints.
Conditions during the incubation and nestling periods of reproduction by birds may have significant effects on offspring growth and development (Drent 1975; Deeming, Rowlett & Simkiss 1987; Webb 1987), but the regulation of conditions within acceptable limits may be energetically costly for parents (Pearson 1994; Williams 1996). The structure of nests may mitigate this energetic demand on parents (Collias & Collias 1984; Hansell 2000). Well-insulated nests slow the rate of egg cooling when an incubating parent departs, thereby reducing the energy required for reheating and maintaining eggs at incubation temperatures (White & Kinney 1974; Drent 1975; Reid, Monaghan & Ruxton 2000a). These energy savings may be reallocated to other phases of reproduction (Reid, Monaghan & Ruxton 2000b). Poorly insulated nests may affect reproductive success by prolonging the incubation period and reducing nestling growth rates (Winkler 1993; Lombardo et al. 1995). However, the construction of a thermally optimal nest must be traded off against building costs, so variation in nest location, structure and properties may result if parents adjust nest construction to suit prevailing conditions (Schaefer 1976, 1980; Kern 1984; Kern & van Riper 1984). Several studies have investigated spatial variation in the insulating properties of nests across a geographical range, but changing climatic conditions through a breeding season may also result in temporal variation in nest structure and function (Reid et al. 2002).
In this paper, we investigate the structure and function of nests of the Long-Tailed Tit, Aegithalos caudatus. Long-Tailed Tits build elaborate domed nests, lined with up to 2600 feathers (Lack & Lack 1958; Gaston 1973; Hansell 1993, 2000). Pairs are single-brooded, but repeat nests are common (Hatchwell et al. 1999). Repeat nests are built faster than first nests (Gaston 1973), and Riehm (1970) found that they contained fewer feathers and were less compact. The aim of this study was to investigate the causes and consequences of variation in the insulating properties of Long-Tailed Tit nests. First, we quantify seasonal changes in the structure of nests and determine which variables influence their insulating properties. We show that the feather lining affects nest insulation quality and that there is a seasonal decline in the mass of feather lining. Secondly, we describe an experiment to test three hypotheses for the seasonal decline of feather mass: (1) feather mass declines because breeders match insulating properties to environmental conditions, so as ambient temperature increases feather mass drops; (2) feather mass declines because late in the season feathers are limiting; and (3) feather mass declines due to time constraints; if collecting feathers is costly (Hansell 2000) it should be traded off against the benefit of completing breeding early in the season (MacColl & Hatchwell 2002; Hatchwell et al. 2004). The experiment involved the addition of feathers to nests. If feather mass is matched to environmental conditions we predicted that supply of extra feathers would not increase the total feather mass of nests relative to controls. However, if feathers are limiting, feather mass would be higher in nests provided with feathers compared with control nests. Finally, if feather mass declines because of time constraints, the interval between the start of lining and laying would be shorter for nests given feathers compared with controls.
Materials and methods
A colour-ringed population of 32–84 pairs of Long-Tailed Tits was studied from 2000 to 2002 in Melton Wood (53°31′N, 1°13′W), Doncaster, UK. Nests were found during the building stage and visited every 1–2 days until they were depredated or until broods fledged. The first nest built by a pair in a breeding season was termed the first nest and subsequent nests by the same pair in the same breeding season were classified as second nests (few pairs built more than two complete nests in a season). Both sexes build the nest and there are two construction phases: building of the outer structure and lining of the interior with feathers, termed ‘feathering’ hereafter (Brown 1924). The outer structure comprises various materials including moss, plant fibres, spiders silk and lichen flakes (Hansell 1993, 2000; A. McGowan, S. P. Sharp & B. J. Hatchwell, personal observation). The outer structure of first nests took an average of 22·7 ± 1·17 SE days (n = 30 nests found at the very start of building) to complete, followed by 15·6 ± 1·25 SE days (n = 25 nests) of feathering by both birds. However, second nests are built much quicker (11·2 ± 1·88 days, n = 11) than first nests (38·3 ± 1·55 days, n = 26; Mann–Whitney U-test, z = 4·87, P < 0·001).
In 2001, we measured the height of the entrance hole from the ground (‘nest height’) and the direction that the nest hole faced (‘nest bearing’) of 41 nests. Nest contents were checked by touch to determine the start of laying and nests were assumed to be complete when the first egg appeared. On the day the first egg was laid, cooling trials were conducted to determine ‘nest insulation quality’ (the egg was removed during trials). Two ellipsoid blocks of hardened Fimo modelling clay (Eberhart, Neumarkt, Germany) were submersed in water at 90 °C, and left until the water temperature had fallen to 75 °C. Both blocks were then removed, dried and a thermistor (PB-3221-S10, Gemini Data Loggers UK Ltd, Chichester, West Sussex, UK) connected to a Tinytag Plus datalogger (TGP-0073, Gemini Dataloggers UK Ltd), was positioned into the centre of each block. One block was placed in the nest while the other was placed just outside the nest at an equivalent height. Both blocks were left to cool for 90 min, dataloggers recording the temperature of each block every 20 s. Trials showed that this protocol allowed both blocks to reach ambient temperature during cooling. Cooling rates were obtained for each block by fitting an exponential equation to the resultant temperature traces:
Temperature of block = ambient temperature + [B exp (−C × time)],(eqn 1)
where B and C are fitted positive constants. B is the initial gradient between the block and the ambient temperature, and C describes the rate of cooling of a block. Nest insulation quality was calculated as the value of C of the nest block minus the value of C of its control block; a large positive difference reflects a high value of nest insulation quality. The cooling rates of the two blocks did not differ significantly when cooled under the same environmental conditions (paired t-test: t9 = 0·199, P = 0·85), so any differences in cooling rates during trials was attributable to the insulating effect of nests. Under controlled environmental conditions this method of calculating nest insulation quality had a repeatability of 89% (F32,65 = 17·57) (Lessells & Boag 1987).
After breeding, Long-Tailed Tit nests were collected and measured using a set of callipers with the nest in an upright position and the nest hole facing forward. Height was measured from the lowest to the highest point of nest material while width was measured at the base of the entrance hole. Depth was measured from the base of the entrance hole to the back of the nest. ‘Nest volume’ was calculated by assuming nests were ellipsoid in shape:
Nest volume = 4/3 × π × rh × rw × rd,(eqn 2)
where rh = height radius, rw = width radius and rd = depth radius. After being placed in conditions of 15 °C and 70% humidity for at least 72 h to allow moisture content to stabilize, nests were dissected and separated into structural material, feathers and dried faeces. Finally, the mass of each component part was weighed electronically.
In 2002, a feather addition experiment was conducted to investigate the seasonal decline in feather mass. Body or contour feathers from domestic poultry in the size range 20–40 mm were used, the preferred size for Long-Tailed Tits (Riehm 1970). Twenty pairs of first nests were matched for the start of the feathering phase and randomly assigned to a feather addition (treatment) or non-addition (control) group. When more than two nests were available to choose from they were paired for their proximity to one another in order to minimize any habitat effect (mean distance between nest pairs: 208 ± 36·5 m, n = 20). All nests were first nests. Feather addition started at least 48 h after feathering had started. Treatment nests had 2 g of feathers added to the interior per day for 4 consecutive days, giving a total of 8 g of extra feathers per experimental nest; this represents approximately 50% of the expected total feather mass of a nest built at that time of year (see Results). The added feathers were smoothed down to fit nest contours. Nests in the control group were treated in the same way, except for feather addition, to control for repeated nest disturbance. No pairs were observed removing feathers from their nest after additions had taken place and no nests were deserted. The first egg date, clutch size, hatch date, number of nestlings at day 11, average chick weight at day 11, and fledging success was recorded for all nests.
Experimental nests were collected after breeding and their volumes calculated. Those nests depredated by mustelids or small rodents remained structurally intact and so were also collected to maximize sample sizes; nests depredated by corvids were torn apart and could not be used for analysis. Nests were left in controlled conditions of 15 °C and 70% humidity for a minimum of 72 h before cooling rates were determined, as described above, in a 5 °C constant temperature room. Finally, nests were dissected and feather and structural masses measured.
Analyses were conducted using SPSS v.10·1 and S-Plus v.6·1. All tests were two-tailed and means ± 1 SE are presented. Observational data were analysed using General Linear Models (GLM). Non-significant effects were removed from the GLM by stepwise deletion (Crawley 1993). ‘Insulation quality’ was used as the response variable and all variables were included in the starting model as covariates, except nest order (first or second), which was included as a categorical variable. The starting model also included higher-order interaction terms considered biologically meaningful. Model residuals were checked for normality and homoscedasticity at each step of removal.
Thirteen of the initial 41 nests were destroyed by corvids before collection, leaving 28 whole nests for analysis. The mean mass of nests was 28·6 ± 1·4 g (n = 28, range = 15·4–44·6 g), feathers representing 41% (range = 21–52%) and the outer structure 59% (range = 48–79%) of total nest mass. There was a significant negative relationship between a nest's feather mass and the date the first egg was laid (Fig. 1a), but no significant relationship between structural mass and lay date, although mass tended to decline through the season (Fig. 1b).
‘Insulation quality’ was determined for 26 nests owing to technical difficulties at two nests. Feather mass varied from 4·9 g to 18·3 g and had a significant positive effect on nest insulation quality (Table 1; Fig. 2a). There was also a significant interaction between feather mass and nest volume (Table 1): small nests had better insulating properties if they had a larger feather mass while there was no effect of feather mass on the insulation quality of large nests (Fig. 2b). In addition, second nests (0·0207 ± 0·001 °C per 20 s, n = 7) had poorer insulation quality than first nests (0·0270 ± 0·001 °C per 20 s, n = 19; Table 1). Note that all these effects were independent of date, which was non-significant and therefore removed from the final model (Table 1).
Table 1. Results of GLM analysis to determine the physical and biological nest variables that affected nest insulation quality. An asterisk represents an interaction between two variables (adjusted R2 = 0·46). Initially non-significant variables removed from the model were: first egg date, nest height, nest bearing and structural mass
Type III sum of squares
Nest volume * feather mass
feather provisioning experiment
There was no significant effect of feather addition on the time between the onset of feathering and first egg date (Table 2a; 8/40 nests were destroyed by predators before laying, so n = 32). A further 14 nests (7 treatment, 7 control) were destroyed by corvids during the incubation and nestling periods, so nest structure and insulation quality were analysed for the remaining 18 nests. There was no significant difference in the outer structure mass of control and experimental nests (Table 2a). This is not surprising because the experimental treatment started only after nest structures were complete. More importantly, the mass of feathers in treatment and control nests did not differ significantly (Table 2a). Furthermore, there was no significant difference in the nest insulation quality of treatment or control nests (Table 2a).
Table 2. The effect of experimental addition of feathers to Long-Tailed Tit nests on: (a) nest construction and nest insulation quality; and (b) measures of parental effort and reproductive success. Experimental nests either received 8 g of feathers (treatment, n = 20) or received similar levels of disturbance but no feathers (control, n = 20). Means ± SE are shown
Wilcoxon Signed Ranks test
(a) Nest construction
Feathering duration (days)
14·7 ± 0·93
15·3 ± 0·79
Structural mass (g)
16·3 ± 1·91
14·0 ± 1·08
Feather mass (g)
19·6 ± 1·99
18·0 ± 1·93
Nest insulation quality (°C/20 s)
0·023 ± 0·0004
0·024 ± 0·0004
(b) Parental effort and reproductive success
10·0 ± 0·38
9·0 ± 0·44
Number of nestlings at day 11
8·8 ± 0·40
8·5 ± 0·76
Mean nestling weight per nest at day 11 (g)
7·5 ± 0·20
7·3 ± 0·20
Duration of incubation period (days)
15·9 ± 0·14
16·6 ± 0·48
Duration of nestling rearing period (days)
16·5 ± 0·20
16·2 ± 0·48
Of the 18 collected nests, 14 (7 treatment, 7 control) had contained nestlings, and 13 nests (6 treatment, 7 control) had produced fledglings. No significant differences were found between experimental and control nests in clutch size or the number and mass of nestlings (Table 2b). Experimental and control nests did not differ significantly in the duration of the incubation or nestling periods either (Table 2b).
Feather mass was the most important nest component for insulation quality. The mass of feathers in nests declined through the breeding season and first nests had better insulation quality than second nests. In addition, there was a significant interaction between feather mass and nest volume, low-volume nests with more feathers having been better insulated than low-volume nests with fewer feathers. We had no evidence that Long-Tailed Tits further enhanced nest thermal properties by orientating the nest towards the sun, as reported in some bird species (e.g. Ricklefs & Hainsworth 1969; Inouye et al. 1981), including Long-Tailed Tits (Riehm 1970).
The mass of feathers used for lining had a positive effect on nests’ insulating properties. This is not surprising because the insulation qualities of feathers are well established (Collias & Collias 1984), but the effect of nest volume on the relationship between feather mass and nest insulation quality would also be expected. Small nests have a higher surface area to volume ratio than large nests and therefore lose heat at a higher rate. Long-Tailed Tits that build small nests could counteract this effect by lining the nest with a larger number of feathers. Alternatively, birds could counter this effect by initially building a larger nest. However, larger nests did not have as many feathers and were no better insulators than small nests that had a large number of feathers (Fig. 2b).
The seasonal decline in the mass of feathers used to line nests, also reported in an earlier study (Riehm 1970), contrasts with the absence of any significant change in structural mass with date. These two results suggest that Long-Tailed Tits selectively adjust the lining component of their nests. Despite the seasonal decline in feather mass there was no associated reduction in nest insulation quality (Table 1); this is probably attributable to increasing ambient temperatures counteracting the effect of a decreased feather mass. This suggests that Long-Tailed Tits build nests with a microclimate that allows eggs to cool within acceptable limits that balance water loss with the risk of chilling, the key threats to embryo viability (Drent 1975; Deeming et al. 1987; Webb 1987). When controlling for the effects of feathers and nest volume, second nests were of poorer insulation quality than first nests. There was no significant difference in structural mass between first and second nests, but the composition or construction of the outer structure might have some influence on nest insulating properties. We did not measure this directly, but Riehm (1970) found that later nests were less compact than early nests and it is certainly the case that the outer structure of second nests is built far more rapidly than that of first nests. Similar results have been reported in other species (Schaefer 1976, 1980; Kern 1984).
Results of the feather provisioning experiment supported the environment-matching hypothesis. Pairs of Long-Tailed Tits supplied with extra feathers brought approximately 50% fewer feathers themselves so that the total mass of feathers and nest insulating properties did not differ between control and treatment nests. This reduction in feathering effort had no significant effect on the duration of the feathering period, suggesting that the seasonal decline in feather mass is due to Long-Tailed Tits tailoring feather mass to environmental conditions. Furthermore, this result shows that Long-Tailed Tits must be able to accurately gauge the thermal environment within their nest and adjust their nest-building behaviour accordingly. There was no evidence that energetic savings during feathering were reallocated to other phases of reproduction, but it should be noted that samples of experimental and control nests available for these comparisons were small because of nest predators.
Results of the experiment also suggested that feathers were not a limiting resource in our study site. This is consistent with the only previous study that has investigated the availability of feathers for nest-builders, which found that supplied feathers were barely used by a range of passerine species, including Long-Tailed Tits (Hansell & Ruxton 2002). Finally, the experiment indicated that the seasonal reduction in the feather lining of nests was not due to time constraints. However, this hypothesis cannot be ruled out entirely because only first nests were involved in the experiment. If Long-Tailed Tits time completion of first nests to the time when they are physiologically capable of laying eggs, treatment birds may not have been able to lay eggs any earlier even though their nests were complete. Given that productivity declines sharply through the season (MacColl & Hatchwell 2002), time constraints on the construction of late nests are likely and it would be worth repeating this experiment with second nests.
In conclusion, the feather lining of Long-Tailed Tit nests is important in determining a nest's insulating properties; breeders appear able to gauge the thermal environment within a nest and adjust their nest-building behaviour accordingly. This study focused on the thermal environment of nests with regards to egg cooling rates and it would be interesting to extend this approach to consider the effects of nest construction on nest humidity and the thermal environment for nestlings and roosting adults.
We thank Andrew MacColl, Ian Hartley, Tim Birkhead and Dave Hazard for their invaluable input and Doncaster City Council for allowing us to watch birds on their land. We thank Robert Ricklefs and an anonymous referee whose comments helped to improve the manuscript. We also thank the numerous undergraduates who helped in measuring and dissecting nests. AMcG was supported by a NERC studentship.