Low incubation temperature slows the development of cold tolerance in a precocial bird

The demands of avian incubation often parallel, or even exceed, those during nestling feeding (Nord and Williams, 2015). Thus, incubation is a trade-off between parental self-maintenance and investment in keeping the eggs at a temperature that is conducive for growth and maturation (Monaghan and Nager, 1997; Reid et al., 2002). In line with this trade-off, parents incubating under strenuous conditions, such as when food availability is low, when it is cold, or when clutches are large, reduce the amount of heat passed to the eggs through reduced incubation temperatures and/or via changes in the amount of time spent incubating (Ardia et al., 2010; MacDonald et al., 2013; Nord and Nilsson, 2012; Nord et al., 2010). By analogy, parents invest more in keeping eggs warm when the demands of incubation are relieved (e.g. by experimental heating of the nest or clutch; Ardia et al., 2009; Reid et al., 2000). It follows that average incubation temperature in the wild varies depending on both environmental factors and parental conditions.

Even slight variation in egg temperature may have far-reaching effects on offspring phenotype (DuRant et al., 2013b). Accordingly, chicks from eggs that develop at low average temperature hatch with reduced energy reserves (DuRant et al., 2011b; Hepp et al., 2006; Olson et al., 2006), have a lower growth rate (DuRant et al., 2010), are smaller at independence (Nord and Nilsson, 2011), and show reduced locomotor performance (Hopkins et al., 2011), lower immunocompetence (DuRant et al., 2011a) and increased stress sensitivity (DuRant et al., 2010) compared with chicks from eggs that develop at high temperatures. These effects can be large enough to reduce long-term survival (Berntsen and Bech, 2016; Hepp and Kennamer, 2012; but see Nord and Nilsson, 2016).

Low incubation temperature can increase the metabolic rate of chicks (e.g. Nord and Nilsson, 2011). This has been interpreted as a maternal effect that prepares chicks for life in a cold world, on the premise that high metabolic rate is indicative of improved thermogenic capacity (Nichelmann and Tzschentke, 2002). Pre-natal temperature conditioning in this manner is exploited by the poultry industry, where short-duration (≤24 h) thermal stimuli from the second trimester (when the hypothalamus–thyroid–pituitary–adrenal axis starts to develop) improve post-hatching cold or heat tolerance depending on the direction of the manipulation (Shinder et al., 2011; Yahav, 2009). However, the ecological relevance of these studies is unclear, as free-ranging birds are unlikely to experience such precisely timed and dosed temperature variation (Nord and Giroud, 2020). Temperature variation in nature is likely to be more constant, e.g. for the duration of a developmental stage, in line with consistent differences in parental investment. The phenotypic consequences of this long-term exposure are often markedly different from those following short-term manipulation in poultry studies. For example, continuous mildly hypothermic incubation reduces chicks’ tolerance of long-term cold exposure, whereas continuous mildly hyperthermic incubation increases it (DuRant et al., 2013a, 2012). However, we are still missing studies that assess how cold- and warm-incubated chicks deal with low temperature during shorter time periods, such as when precocial offspring alternate short foraging bouts with brooding by parents. This is of great ecological interest, because low body temperature (T_b) will limit feeding bout duration (Pedersen and Steen, 1979) and, hence, body mass gain (Jørgensen and Blix, 1985).

We investigated whether variation in incubation temperature alters thermoregulation over short time periods, similar to feeding bouts in the wild. We incubated Japanese quail (Coturnix japonica) eggs at three biologically relevant temperatures and then measured chicks' capacity to resist cooling in a common garden setting during the first week of life. If low incubation temperature is conducive for subsequent cold tolerance (as in poultry studies), we predicted that cold-incubated chicks would show increased cold tolerance. If, in contrast, continuously sub-normal incubation temperature constrains growth and maturation (as in studies of wild birds), we predicted that cold tolerance would be higher in warm- than in cold-incubated chicks.

Japanese quail, Coturnix japonica Temminck and Schlegel 1849, eggs (Stjärnås Djur och Fjäder, Lessebo, Sweden) of unknown sex were incubated in three incubators (Ruvmax, Ödskölt, Sweden) maintained at low (35.5°C; 19 eggs), normal (37.0°C; 20 eggs) or high (38.5°C; 20 eggs) temperature. The incubators were set up next to each other in a room with constant photic and thermal conditions. Our previous work on another model, but with the same incubators, has shown that the incubator per se does not influence the phenotypic effects of variation in developmental temperature observed in chicks (Nord and Nilsson, 2011). We measured temperature (±0.0625°C) immediately adjacent to the eggs in 24 h cycles, using a temperature data logger (iButton DS1922-L, Maxim, Sunnyvale, CA, USA; accuracy ±0.5°C), to ensure that the eggs were at target temperature. Relative humidity in the incubators was maintained at 65–70% until pipping, and was then increased to 80%.

At hatching, we weighed the chicks to the nearest 0.1 g and marked them uniquely on the tarsi using non-toxic felt-tipped pens. They were then transferred to a holding pen (1.5×2.0 m) with a bedding of Populus wood shavings and ad libitum access to food (poultry chick feed; Lantmännen, Malmö, Sweden), water and a heat lamp (38.5°C at floor level). Air temperature in the facility ranged from 15 to 17°C.

All procedures were approved by the Malmö/Lund Animal Ethics Committee (permit no. M 238-07).

The cooling experiment was designed to simulate chicks alternating between short foraging bouts and parental brooding in the wild. Thus, starting one day after hatching (henceforth day 1), we collected chicks from the pen and immediately (within 10 s of capture) measured T_b by inserting a type K thermocouple (diameter: 0.9 mm) connected to a Testo 925 thermometer (Testo AG, Lenzkirch, Germany) 15 mm through the cloaca (insertion to further depths did not alter the readings). Less than 30 s later, the chicks were put in open 1 l plastic vials inside a climate chamber (BK600, Vötsch Industrietechnik, Balingen, Germany) kept at 20°C for 10 min, after which we measured T_b again, weighed the chick, and returned it to the pen. This was repeated every other day until day 7, when it was assumed that the chicks had reached homeothermy (as defined by Ricklefs, 1987; Visser, 1998).

Hatching success varied between treatments, being highest at 37.5°C (70%; 14 of 20 eggs), lower at 35.5°C (53%; 10 of 19 eggs), and lower still at 38.5°C (40%; 8 of 20 eggs). There were no visible signs of development in the eggs that did not hatch. We did not record any mortality in the two higher temperature treatments, but four chicks in the 35.5°C group died between day 1 and day 3, and one died between day 3 and day 5. Hence, sample sizes differed with age in this treatment.

We used R 3.6.1 (http://www.R-project.org/) for the statistics. We compared the length of the incubation period and body mass at hatching using linear models (lm function in R base) with treatment as a factor. Body mass, R_ΔTb and H during the experiment were compared using linear mixed effects models (lmer function in lme4; Bates et al., 2015) with treatment, day and treatment×day as factors and chick identity as a random intercept. The interaction was removed when non-significant, but all main effects were retained. P-values for the mixed models were inferred from likelihood ratio tests. Parameter estimates and their s.e.m. were calculated using the emmeans package (Lenth, 2016). When the interaction was significant (P<0.05), post hoc tests were performed between treatments within days. Data in the text are predicted means±s.e.m.

The length of the incubation period was temperature dependent, being shortest at 38.5°C (16.0±0.1 days), longer at 37.0°C (17.3±0.1 days) and markedly prolonged at 35.5°C (19.7±0.1 days) (Fig. 1A, Table 1). At hatching, chicks incubated at 38.5°C (8.59±0.20 g) and 37.0°C (8.15±0.15 g) weighed significantly more than chicks incubated at 35.5°C (7.46±0.18 g) (Fig. 1B, Table 1).

The body mass gain during the first week of life differed between the treatments (treatment×age: P<0.001) (Fig. 2A, Table 1). High-temperature chicks were heavier than chicks from the two other groups on day 1. Chicks incubated at 37.0°C subsequently recovered, such that body mass was similar to that in the 38.5°C group for the remainder of the experiment, and well above the body mass of 35.5°C chicks (Fig. 2A).

Chicks from the 35.5°C group suffered a larger reduction in T_b during the cooling challenge than chicks from the two other groups throughout the experiment (Fig. 2B). When this was expressed as R_ΔTb (Eqn 1), we found that cold tolerance improved with age at a similar rate in all groups (Fig. 2C, Table 1). However, chicks that had been incubated at the two highest temperatures had lower cooling rates than those incubated at 35.5°C throughout the study (Fig. 2C, Table 1).

H (Eqn 2) improved with age by a similar magnitude in all groups (Fig. 2D, Table 1). Yet, cold-incubated chicks were less homeothermic than chicks from the other two treatments during all measurements (Fig. 2D, Table 1).

We found that low incubation temperature slowed development both before (Fig. 1) and after hatching (Fig. 2A), and that these effects were large enough to affect the capacity to withstand cooling representative of that during a typical foraging bout of a precocial chick (Fig. 2B–D). This corroborates research on wild precocial birds where the cooling challenge was substantially longer and stronger than in this study (DuRant et al., 2013a, 2012). It is interesting to note that developmental trajectories for high- and mid-temperature chicks were rather similar, in line with results obtained for a range of traits in other bird studies (e.g. DuRant et al., 2010, 2011a,b; Hopkins et al., 2011; Nord and Nilsson, 2011). This highlights that effects of incubation temperature on phenotype are non-linear. Here, this non-linearity was already evident at the embryonic stage: a 1.5°C temperature increase advanced hatching by 1 day with no change in hatchling size, but a corresponding decrease prolonged incubation by 3 days and resulted in smaller hatchlings.

Because we did not measure any thermogenic responses to cooling, i.e. metabolic heat production and shivering, we do not know whether reduced cold tolerance in chicks from the 35.5°C group was a reflection of increased dry heat transfer on account of a larger surface area to volume ratio, a dampened or delayed thermogenic response to cooling, or both. It is unlikely that the response was caused by size-related differences in heat loss alone, because precocial chicks produce heat in response to cold a day after hatching (Marjoniemi and Hohtola, 1999), with a transition from a weak to a pronounced thermogenic response a few days later when skeletal muscles grow and mature (Aulie, 1976). It has been suggested that the purpose of this early life facultative heat production is not to maintain T_b per se – probably a futile endeavour in light of the low thermal mass of young chicks – but rather to reduce the rate of body cooling to make feeding time less dependent on environmental temperature (Jørgensen and Blix, 1988). In view of this, it is tempting to speculate that the observed body cooling was not a reflection of low thermogenic capacity but rather an adaptive response to conserve energy whilst maintaining some foraging capacity. Even so, it seems likely that chicks incubated at higher temperature would be able to extend foraging time in early life more than chicks incubated at lower temperature, because their larger size should contribute directly to both thermogenesis (as there is more heat-producing tissue) and insulation (as thermal properties are more conducive for heat retention in a larger body). This is well in line with the observation that it took chicks from the 35.5°C treatment 2 days longer (i.e. until body mass was comparable) to reach the cooling rate and degree of homeothermy that mid- and high-temperature chicks had already attained by day 3 (Fig. 2).

In the wild, the observed physiological and biophysical effects could exacerbate other consequences of low incubation temperature, e.g. smaller size, higher energy turnover rate and reduced locomotor performance (DuRant et al., 2010, 2011b; Hepp et al., 2006; Nord and Nilsson, 2011), because precocial chicks with reduced capacity to withstand cooling probably must reduce foraging bout length to avoid costs of hypothermia (Carr and Lima, 2013; Pedersen and Steen, 1979). Shorter bout length, together with any reduction in foraging efficiency from hypothermia-related cognitive impairment (Rashotte et al., 1998), could hamper total food intake with downstream consequences for growth and survival (Jørgensen and Blix, 1985). Thus, growth in cold-incubated chicks might suffer not only from shorter foraging bouts but also by reduced energy acquisition rate during that time. From the perspective of the female, it is interesting to speculate that investment in self-maintenance by reduced incubation effort (cf. Nord and Williams, 2015) could result in a cost later during the same breeding event, if a parent to cooling-intolerant chicks has to divert time from self-feeding to brooding.

We have demonstrated that even slight changes in the thermal environment that eggs experience can have sufficiently large effects on growth and maturation to impact how well chicks withstand cooling. Future studies should elucidate the proximate nature of this observation, and also assess how thermoregulatory consequences of low incubation temperature impact the foraging behaviour of young chicks, and the brooding behaviour of their parents, in wild models of precocial birds.

Stefhan Stjärnås kindly provided quail eggs for the study.