In birds ambient temperature can influence adult incubation behaviour, energy budget, egg temperature, and embryonic development with downstream effects on offspring survival. Surprisingly, experimental manipulations of the whole nesting environment to test causes and consequences of variation in incubation pattern, energy balance, egg temperature, and the duration of development are lacking to date. Here, we bred pairs of Zebra Finches Taeniopygia guttataunder controlled conditions at 18°and 30°C and measured clutch size, temperature, hatching success, parental attentiveness and the length of the embryonic period. We found that when breeding at the higher temperature, males, but not females, increased the number of incubation bouts on the nest. Instead, females, but not males, reduced their attentiveness towards the clutch overall. Eggs showed no temperature differences between the two treatments and bigger clutches experienced lower temperatures. This suggests that parental behaviour may buffer the effect of ambient conditions on the thermal profile of eggs, including species with high rates of parental attentiveness. Warmer conditions yielded higher hatching rates but did not cause measurable differences in the length of embryonic development. Still, smaller clutches hatched earlier in accordance with the higher temperature experienced. Additionally, we used data from the literature to calculate parental energy expenditure and demonstrate that this was substantially different across the two treatments, although predicted energy savings from reduced attentiveness at 30°C appeared negligible. These results suggest that when food is available, ambient temperature and not energy trade-offs may explain variation in incubation behaviour. 

Experimental design

Twenty-three male and female zebra finches (domestically bred) were force-paired and reared in captivity under controlled conditions at Macquarie University (under Animal Ethics Committee approval ARA 2013/029). Each pair was housed in a cage with dimensions of 70 cm wide × 47 cm deep × 130 cm high containing two 13.5 × 15 cm rattan nest baskets and provided with November Grass Amphybromusspp., white cotton thread and Emu Dromaius novaehollandiaefeathers to line their cups. Cages were held inside temperature-controlled rooms set at either 18°C or 30°C (n= 12 pairs/temperature) where finches were acclimated for two weeks without nests before being allowed to breed. These two treatment temperatures are ecologically relevant because they reflect the average ambient temperatures recorded under natural conditions at Fowlers Gap Arid Research Station (our study field in Australia 31° S, 141° E), during the early cooler (July-August), and late hotter (December-Jannuary) months of the breeding season (Griffithet al.2016). Birds were maintained with dry millet finch seed mix (Panicum and Setaria spp.) and water available ad libitumplus a small daily supplement of Green Pea (Pisum sativum)-Spinach (Spinacia oleracea) mash with soft food, hard boiled Chicken (Gallus gallus domesticus) egg, and sprouted seed. During the egg-laying stage of the first reproductive attempt nests were monitored daily, with any new eggs collected for other work (Andrew et al. 2018), and replaced with dummy eggs (made from white Mont Marte hardening modeling clay, Mont Marte, Australia) until clutch completion. This first attempt was interrupted on day 9 of incubation when nests were removed for another study looking at temperature-dependent variation in nest size and building material (Campbell et al.2018). At this point, new rattan nesting baskets and material were provided and birds were allowed a second attempt where they could rebuild their nest, lay and incubate a new clutch, and raise a brood of chicks to independence.Pairs were all then switched into the alternative temperature treatment and following another two-week acclimation period with no nest baskets or material, the entire process was repeated(seeAndrew et al.2017for further methodological details). As a result, each pair had two reproductive attempts and completed a full successful breeding cycle (from nest building to fledging) in the two different temperature treatments. In this way, over the experimental period, half of the pairs bred first at 18°C then 30°C and the other half bred first at 30°C then 18°C. This design allowed us to take a within-individual analytical approach, reducing the confounding effects of differences across individuals, and also the effects of the first and second breeding attempt by each pair. 

Measurements of parental behaviour, clutch temperature and embryonic period length

Between days 5 and 7 of the incubation of dummy eggs, average egg temperature was quantified using an iButton (DS1921G Thermochron iButton: Maxim Integrated, San Jose, CA) set to record temperature (±0.1°C) every minute. The device was placed in the bottom of the nest cup, alongside the eggs before 09:00 (GMT+10) in the morning and removed at the same time two days later. We decided to measure temperature in nests with dummy eggs for two main reasons. The first was to prevent the iButton from damaging the actual eggs. The second was to avoid increasing the total mass in the real nest by adding the iButton. Indeed this could dissipate the amount of heat received by the eggs and influence timing of embryonic development as found in other studies (Moreno & Carlson 1989; Thompson et al. 1998). The morning following the first day of egg temperature measurements we video recorded parental incubation behaviour at the nest for six hours starting between 08:00 and 09:00 am. Overall, the majority of videos were recorded on day 7 of incubation, with seven occurring on day 8. The total video duration of six hours was split in three two-hour sections called early, mid and late morning, respectively. One random hour per section was watched thereby providing three hours of parental incubation attentiveness for each pair. Measuring clutch temperature with the iButton while also video recording parental behaviour allowed us to match the timing of parental on and off bouts with the temperature recorded in the nest. The temperature data recorded were then averaged to obtain mean incubation temperature for each sex. This process was repeated for each randomly selected one-hour interval of incubation (early, mid and late morning) and for both breeding cycles. Therefore we ended up analyzing a dataset of 276 data points generated from 23 pairs × two sexes  (male or female) × three distinct part of the day (early, mid and late morning) × two treatment temperature, × two breeding cycles that yielded 276 observations (see complete data set for temperature and attentiveness analysis Dryad). Because Zebra Finches in captivity start incubating as soon as they lay eggs, rather than at clutch completion (Gilby et al. 2013), the duration of the embryonic period was calculated as the number of days between the laying and the hatch date of the first egg in the clutch. To capture hatch date nests were monitored every day only in the morning between 07:00 and 8:00 to avoid interferences with other tests and measurements performed during the day. Timing of hatch was assigned to the previous or same day based on presence of eggshell fragments in the nest, appearance of the nestling (wet versus dry), and mass of the nestling compared to the mass of the eggs during late incubation. Note that the developmental time was measured for the second reproductive attempt at each temperature treatment, while the egg temperatures and behavioural videos were taken while the parents incubated dummy eggs that replaced the first clutch that was laid. None of the pairs we used in our experiment died so that we could use the same adults in both breeding cycles at two different temperatures. 

Data analysis

We tested for the effects of the temperature treatment on the amount of time spent in the nest, and on number of incubation bouts by fitting a linear mixed model with treatment (18°or 30°C), breeding cycle (first or second), order of exposure to treatment (18à30 or 30à18°C), and sex (male or female) as independent variables. Individual pair identity was included as a random factor. We also built a model to examine differences between temperature treatments on clutch temperature and percent of eggs hatched with treatment, sex, breeding cycle, clutch size and order as independent variables and pair identity as random factor. Finally, we evaluated the effect of experimental temperature on length of embryonic period using treatment, breeding cycle, clutch size, and order as independent variables and pair identity as a random factor. Marginal and conditional R²for our generalized mixed-effects models were calculated using the r squaredGLMM function MuMIn that implements the method described by Nakagawa and Schielzeth (2013). We tested for two- and three-way interactions between all independent variables included in each model. We dropped all interactions when not significant. Descriptive statistics (mean andSD) are reported in the Results section while model outputs are summarized in the tables. We used a threshold α = 0.05 to test the significance of our models. All statistical analyses were performed using packages lme4 and in program R version 3.4.2 for Mac (R Development Core Team 2008). P-valueswere obtained using package lmerTest (Kuznetsova et al. 2016).All data and codes used in our study are available in Dryad and in Appendix S1 of Supporting Information, respectively).

            To quantify the differences in energy expenditure (kJ) between incubating and non incubating birds at 30°C and between birds breeding at the two treatment temperatures we used the equations provided by Vleck (1981). To translate the amount of energy calculated from those equations into equivalent grammes of food we used the calculations provided by Harper et al. (1998) (see Supporting Information Appendix S2 for details of how we obtained our results).