Studies have shown that organisms may adjust metabolic rates in response to thermal variability, highlighting metabolic plasticity as a key adaptive mechanism. Understanding the extent of metabolic plasticity of an organism is key to predicting its adaptation to climate change. In birds, the embryos from the uniparentally incubating species are more frequently exposed to intermittent cooling due to the parents’ repeated absences from the nest than the embryos from the species with biparental incubation. Such pressure may favour them to evolve a better ability to cope with the physiologically suppressive effects of low temperatures and temperature fluctuations. We compared embryonic heart rate, a proxy for embryonic metabolic rate, and its response to egg temperature change between two closely-related species, the biparentally incubating Black-throated Tit (Aegithalos concinnus) and the uniparentally incubating Silver-throated Tit (A. glaucogularis). We also conducted an interspecific egg-swapping experiment to investigate the effect of incubation environment on embryonic metabolism and its consequences (i.e., incubation period length and hatching success). Consistent with the expectation, Silver-throated Tit exhibited a trend (although nonsignificant) of higher embryonic heart rate than Black-throated Tit. Also, when egg temperatures dropped, Silver-throated Tit showed a significantly slighter decrease in embryonic heart rate than Black-throated Tit, suggesting that they may better cope with temperature drop. In the egg-swapping experiment, embryonic heart rates did not differ significantly between fostered and unfostered eggs, but following fostering, heart rates tended to decrease in Silver-throated Tit embryos and increase in Black-throated Tit embryos, suggesting that embryos of both species show some responsiveness to new thermal environments. Egg-fostering treatment did not affect hatching success in either species, but the fostered eggs of Silver-throated Tits and Black-throated Tits tended to shorten and lengthen their incubation period, respectively, implying that the relatively stable biparental incubation environment may facilitate faster embryonic development than the more variable uniparental incubation environment. The findings enhance our understanding of the developmental strategies and responsiveness to changed environments in avian embryos under different parental care modes and provide insights into their potential to respond to climate change.

Research species and fieldwork

This study was conducted in Zhanglou Village of Luoshan County, which is near the Dongzhai National Nature Reserve of Henan Province, China (31.95° N, 114.25° E). The two species, Black-throated Tit and Silver-throated Tit, occur sympatrically in this region and breed synchronously. They typically begin nest-building in late January or early February. Their nests are dome-shaped and similar in size, although slightly different in outer-layer nest materials (Li et al., 2012). Egg-laying generally starts in late February or early March, with one egg laid per day. Clutch sizes typically range from 4 to 8 eggs in Black-throated Tits and 5 to 9 eggs in Silver-throated Tits (Hu et al., 2024), with both species most frequently laying 7-egg clutches (Li et al., 2012). The average egg mass (mean ± SE) of Silver-throated Tits (0.85 ± 0.01 g) is heavier than that of Black-throated Tits (0.72 ± 0.01 g) (Li et al., 2012).

Both species usually initiate incubation on the day the clutch is completed, though in the late breeding season, incubation may begin prior to clutch completion. The two species did not differ significantly in average nest temperature (Hu et al., 2024), but nests of Silver-throated Tits experienced more frequent nest temperature fluctuations than nests of Black-throated Tits (mean number of fluctuations: 15.9 vs. 4.8 during the daytime; Hu et al., 2025a). The incubation period (mean ± SD) was 12.62 ± 1.01 days in Black-throated Tit and 13.85 ± 1.18 days in Silver-throated Tit (Hu et al., 2024). Major nest predators during the breeding season included corvids, Asian Barred Owlets (Glaucidium cuculoides), snakes, and small mammals (Li et al., 2012). The two species usually produce one successful brood each year but may renest after a nest failure.

The Black-throated Tit and Silver-throated Tit populations at the study site have been monitored since 2011. Birds were captured using mist nets either during the winters (typically December to January) or breeding seasons (typically February to May). Each bird was banded with a unique combination of coloured rings and a metal ring for individual identification. During the breeding seasons, nests were located by searching or by following adult birds and then monitored regularly. Nests were usually checked every three days during nest-building, every two days during the egg-laying and incubation period (but more frequently when laying was imminent), and daily near hatching to determine hatch dates. For additional information on their breeding ecology and general field procedures, see Guan et al. (2018), Hu et al. (2024), and Li et al. (2012).

Egg-swapping experiments

Egg-swapping experiments were conducted during 2019 and from 2022 to 2024. In addition to routine nest monitoring, nests were checked daily between the laying of the 6^th egg and clutch completion to determine the onset of incubation. This was assessed by the female incubation behaviour and by touching eggs, as eggs in nests not yet incubated were obviously cool and exhibited a temperature contrast with incubated eggs (based on researchers’ experience). Nests from the two species where incubation initiation dates differed by no more than one day and had similar clutch sizes (± 1 egg, within the natural clutch size range of both species) were paired, prioritizing the nests with close proximity. One exception included a pair with 9 eggs in the Silver-throated Tit and 7 eggs in the Black-throated Tit.

Entire clutches were cross-fostered between paired nests on the first day of incubation (19 Black-throated Tit and 16 Silver-throated Tit nests) or the second day (24 Black-throated Tit and 27 Silver-throated Tit nests). Fostered eggs were marked to distinguish species. In 2019, eggs were returned to their natal nests on day 9 or 10 of incubation. In 2022 ~ 2024, to prolong embryos’ exposure to a fostering nest environment, eggs were returned to their natal nests on day 10 or 11. No embryos hatched before the eggs were returned to their natal nests, except in cases where nests failed. During all transfers, eggs (including those for embryonic heart rate measurements, see below) were transported in cotton-lined plastic boxes heated with warm water bottles or heat packs, which were pre-tested for temperatures not exceeding 40 °C.

Each egg transfer session lasted 30 mins. Given their high natural nest failure rate (e.g., 76% for Silver-throated Tit; Guan et al., 2018), wire meshes were placed surrounding the experimental nests to protect them from predation until nestling hatching. These cages (mesh grid size: 3 × 3 cm) allowed access by Black-throated Tis and Silver-throated Tits but deterred some predators ~(~e.g. corvids). Despite the protection, 23 out of 86 paired nests (26.7%) failed during incubation.

Experimental nests that underwent egg swapping were classified as the experimental group, while naturally incubated nests served as the control group. To ensure their temporal comparability, only natural nests with the date laying the first egg falling within the earliest and latest first-egg-laying dates of a year’s experimental nests of each species were included as controls.

Embryonic heart rate and egg temperature measurements

The embryonic heart rate measurements were conducted in 2023 and 2024. Considering the substantial workload involved in measuring entire clutches, only a subset of eggs was selected for embryonic heart rate measurement. As differences in embryonic heart rate may exist between early- and late-laid eggs within a clutch (Boonstra et al., 2010; Cones & Westneat, 2024; Wang et al., 2023), the 3^rd to 5^th egg in each nest was chosen for measurement. Nests were checked every two days, and once two eggs had been laid, daily marking of each newly laid egg continued until the 5^th egg was laid, to distinguish the 3^rd, 4^th, and 5^th eggs.

Besides, due to limited existing evidence, it remains unclear at which embryonic developmental stages a changed incubation environment is most likely to influence metabolic responses. In addition, the metabolic responses to environmental variation are energetically costly (DeWitt et al., 1998; Monaghan et al., 2009), so embryos may have sufficient energetic capacity to respond to thermal variation early in development, but the capacity may be reduced at a later development stage if energy reserves have already been consumed (Vleck & Vleck, 1987). As a result, responsiveness to temperature may differ between early and late incubation stages. For these reasons, we collected and analysed data separately for early and late incubation stages (see below). This approach allowed us to explore whether responsiveness to incubation environment is stage dependent, i.e., whether the relationships between temperature and heart rate differ across developmental stages of embryos.

Previous studies have shown that in small passerines such as Zebra Finches, embryonic heart rate usually becomes detectable on day 5 of incubation (Sheldon et al., 2018). To collect data from both early and late developmental stages of Black-throated Tit and Silver-throated Tit’s embryos, we measured embryonic heart rates on day 5 and day 10 after the start of incubation, with the measurement on day 5 representing the early developmental stage, and that on day 10 representing the late stage. It is worth noting that among the 39 Silver-throated Tit nests and 54 Black-throated Tit nests in which embryonic heart rates were measured, one Silver-throated Tit nest was measured on day 6 and day 11 of incubation, and one Black-throated Tit nest was measured on day 9. The measurement on day 6 was treated as the early stage, while the measurements on days 9 and 11 were considered as the late stage.

When transferring target eggs for embryonic heart rate measurement, failed eggs from a deserted nest or paraffin eggs of similar size were placed in the nest to replace the removed eggs, maintaining at least five eggs per nest to prevent nest abandonment. The dummy eggs were then removed when the target eggs were returned. For each measured egg, the nest of origin, whether it came from a fostered clutch, and its developmental stage (early or late) were recorded. Each egg was then transferred to an indoor incubator (Yike·Beite Incubator, LN2-16), preheated to ~38 °C, and maintained at 50 ~ 70% humidity. After warming in the incubator for at least 15 mins, eggs were randomly selected and removed one at a time for embryonic heart rate measurement using a Buddy MK2 Digital Egg Monitor (Vetronic Services, UK; Sheldon et al., 2018). We attempted to measure embryonic heart rate at 1-minute intervals. However, due to the small size of the embryos and their movement within the egg (Sheldon et al., 2018), stable detection was often delayed or quickly lost. When embryonic heart rate became undetectable, the egg’s position was gently adjusted (Sheldon et al., 2018) until a stable signal was obtained. Thus, actual intervals between measurements were often longer than one minute. The average time from removal from the incubator to the first successful stable embryonic heart rate reading was 43.47 ± 49.59 s (mean ± SD), ranging from 4 to 517 s (n = 358). Once the embryonic heart rate stabilized, the value was recorded, and the egg’s surface temperature was immediately measured using a thermal imaging camera (HIKMICRO-H10, accuracy: ± 2 °C) at a distance of 15 cm, with the highest surface temperature recorded as the egg temperature. Multiple embryonic heart rate readings were taken for each egg until its highest surface temperature dropped to ~27 °C. After measurements were completed, the egg was returned to the incubator for reheating until all eggs of the clutch were measured and could be returned to their nest.

We did not choose to measure embryonic heart rate immediately after eggs were removed from the nest. This was because the nest entrances were small and the process of egg removal inevitably took some time, leading to a decrease in egg temperature. As a result, it was not easy to obtain embryonic heart rate values at high egg temperatures in the field. Measuring eggs indoors after re-heating ensured that we could capture embryonic heart rate across the full range of ecologically relevant incubation temperatures. While transport and re-heating may introduce some stress or transient effects, all eggs were subject to the same procedures under standardized conditions.

On average, 5.43 ± 3.07 (mean ± SD) embryonic heart rate values were recorded for each egg at the two developmental stages (range: 1 ~ 11, n = 420). After embryonic heart rate measurement, each egg was weighed to the nearest ± 0.001 g. The time from egg removal from the nest to being returned on the same day averaged 1.96 ± 0.71 h (mean ± SD; range: 0.90 ~ 4.33 h, n = 146). However, due to the large number of nests requiring measurement on certain days, some eggs were not returned on the same day (involving 8 Black-throated Tit nests and 7 Silver-throated Tit nests; see Table 2 for details). In such cases, the eggs were kept in the incubator overnight with the temperature set to ~37.8 °C and humidity maintained at 50 ~ 70% and were manually turned 1 ~ 2 times between 7 PM and 12 AM. They were returned to the nest the following morning. According to the continuous monitoring, incubation in the incubator for one night did not noticeably affect the hatching success of the eggs.

Ambient temperature data

To monitor ambient temperature, a temperature logger (GSP958, accuracy: ± 0.5 °C; Guangzhou Lexiang Electronics Co., Ltd., China) was placed at each of three major types of habitats favoured by the two species within the study area. Each logger recorded temperature at 1-minute intervals and was replaced every 21 days due to memory limitations. The average of the three loggers was used as the ambient temperature for the study area. However, due to setup issues, some loggers failed to record for 21 days. In these cases, the average of the remaining one or two loggers was used instead. Additionally, technical issues of the loggers occasionally caused anomalous values at specific temperatures (e.g., ~11.9 °C, 25 °C, and 38.1 °C). These anomalies were replaced by the average of the values recorded 1 minute before and after.

To examine the effect of ambient temperature on embryonic heart rate (see below), the average daily ambient temperature from the start of incubation to the day before embryonic heart rate measurement was calculated for each nest to represent the mean temperature experienced during embryonic development. In addition, the mean daily ambient temperature over the entire incubation period for each nest was also calculated to assess its effects on incubation period length and hatching success (see below).

Data analysis

Although some individuals may renest after previous nest failure within a breeding season, only the data from their first clutch in that year were used in the analyses. For individuals recorded for breeding in multiple years, we prioritized using their data from nests involved in the egg-swapping experiment. If no such nests were available, we used the data from their first-year records. Therefore, no individual parent had repeated data in a given analysis. See Table 3 for additional details of our experimental design. All analyses were conducted in R version 4.5.0 (R Core Team, 2025).

Effects of factors on embryonic heart rate

To compare embryonic heart rate between species and among treatment groups, we categorized eggs into four types: Black-throated Tits’ not-fostered eggs (BTN), Black-throated Tits’ fostered eggs (BTY), Silver-throated Tits’ not-fostered eggs (STN), and Silver-throated Tits’ fostered eggs (STY), with fostered eggs meaning those incubated by another species (see Table 2 for sample sizes).

We performed analyses with linear mixed models (LMMs) using the lme4 package version 1.1-37 (Bates et al., 2025). In the LMMs, embryonic heart rate was treated as the response variable. The initial fixed effects in the model included egg temperature at the time of embryonic heart rate reading, egg mass on the day of measurement, the average daily ambient temperature experienced by the embryo from the start of incubation to the day before measurement, laying date (days from December 31 of the previous year to the date laying the first egg for a nest), egg type (i.e. BTN, BTY, STN and STY), year, developmental stage (i.e. early or late stage). According to our hypotheses and predictions, the initial fixed effects also included the interaction between egg type and egg temperature. Because the sensitivity to environmental variation may differ among species between early and late stages of embryonic development, we further incorporated two-way interactions of developmental stage with egg type and with egg temperature, as well as the three-way interaction among developmental stage, egg type, and egg temperature. We included nest ID and egg ID nested within nest ID as random effects to account for non-independence among eggs within the same nest and repeated measures of the same egg.

Because of a high variance inflation factor (VIF), year was excluded from the fixed effects (see below). Accordingly, the final model was: embryonic heart rate ~ egg temperature + egg mass + average daily ambient temperature experienced by embryos + laying date + egg type + developmental stage + egg type egg temperature + developmental stage egg type + developmental stage egg temperature + developmentalstageg stage egg temperature + (1 | nest ID/egg ID). Because the result of the above model (Supplementary Table S1) indicated a three-way interaction (p < 0.05), we conducted separate analyses for early and late developmental stages to better explore the differences between them. The model formula for the analysis of each specific stage was: embryonic heart rate ~ egg temperature + egg mass + average daily ambient temperature experienced by embryos + laying date + egg type + egg type* *egg temperature + (1 | nest ID/egg ID). The results for each stage are presented in Table 4.

Effects of egg-swapping on incubation period length and hatching success 

The data for the incubation period length included 29 experimental and 39 control Silver-throated Tit nests, and 31 experimental and 56 control Black-throated Tit nests. These nests included the nests where the eggs were used to measure embryonic heart rate but excluded those where eggs hatched in fostering nests during the egg-swapping experiment (see above). In addition, a small proportion of nests, including 3 experimental (10%) Silver-throated Tit nests, and 2 experimental (6%) and 4 control (7%) Black-throated Tit nests, had eggs that experienced one night of artificial incubation (see above).

To assess the effects of egg-swapping on incubation period length, we used incubation period length as the response variable in the linear model. Fixed effects in the model included species, treatment (fostered or not-fostered), average daily ambient temperature during the incubation period, clutch size, year, and laying date. As we were interested in the species-specific effect of treatment, we also included an interaction between species and treatment in the fixed effects. Because of high VIF (see below), the variable “year” was further excluded from the fixed effects. Accordingly, the final model was: incubation period length ~ species + treatment + average daily ambient temperature during the incubation period + clutch size + laying date + species treatment.

For hatching success, the dataset included 22 experimental and 37 control nests for Silver-throated Tits, and 22 experimental and 44 control nests for Black-throated Tits. The analysis was conducted using a generalized linear model, with hatching success (proportion of eggs that hatched per nest) as the response variable. Fixed effects were the same as those in the incubation period analysis, with year also being excluded due to high VIF. The final model was: hatching success ~ species + treatment + average daily ambient temperature during the incubation period + clutch size + laying date + species*treatment. The model with a binomial error distribution was found to be overdispersed, as assessed by the testDispersion function from the DHARMa package version 0.4.7 (Hartig et al., 2024). Therefore, a beta-binomial distribution (Young-Xu & Chan, 2008) with a logit link was used for this model.

To account for potential effects of overnight artificial incubation (see above) on incubation period length and hatching success, we re-ran the two analyses, excluding nests where the measured eggs had experienced overnight artificial incubation. The results (presented in Supplementary Table S2) were consistent with those obtained from the full dataset (Table 5), indicating that overnight artificial incubation did not affect our conclusions.

Before analyses, multicollinearity among predictors was assessed using the VIF calculated via the car package version 3.1-3 (Fox et al., 2024). Across all analyses, factors that had a high VIF (> 5) (Akinwande et al., 2015) were excluded (see above). Variables or factor-level differences were considered statistically significant when p < 0.05. To test for pairwise differences among different levels of categorical predictors (e.g., egg types) or interactions between continuous and categorical variables in the above analyses, we conducted post-hoc comparisons in the emmeans package version 1.11.1 (Lenth, 2025).