Study species and field sites
Cardinals were captured with mist nets at the Miami University Ecology Research Center (39°30′N, 84°45′W), in Southwest Ohio from 16 January to 21 March, 2012 and 2014 (winter) and from 8 August to 4 September, 2012 and 2013 (summer). These times were chosen to ensure that birds were experiencing the challenge of winter cold or were captured following the summer breeding season. All birds were caught between sunrise and early afternoon and release was always at least 1 h before sunset to ensure a successful transition back to the environment or aviary following laboratory procedures described below. Thus, animals were kept in the animal facility for no more than several hours. Animals with implanted temperature loggers were kept for 2 weeks as described below.
All birds were captured under Ohio Department of Natural Resources permit no. 11–152 and US Fish and Wildlife Service permit no. MB158451-1. All animal experimentation was approved by the Institutional Animal Care and Use Committee of Miami University (protocol no. 736) and complied with the regulations of the National Institutes of Health as well as the laws of the United States.
Body composition analysis
We measured total body mass, fat mass and lean mass to determine whether body size and/or composition changed across seasons. Upon capture, cardinals (summer, n = 19; winter, n = 19) were weighed using a digital scale with resolution to 0.01 g (Scout Pro, Pine Brook, NJ, USA). We then determined fat mass and lean mass using MRI-based body composition analysis (EchoMRI-SuperFLEX™, EchoMRI, Houston TX, USA). The instrument was calibrated using the manufacturer-provided calibration vessel. These measurements do not detect feather weight, but do include gut contents. Following these measures, we either measured metabolic rates of birds using respirometry or surgically implanted a temperature logger, as described below.
Respirometry
As neither feather mass nor the contribution of feathers to insulation is quantified by NMR-based body composition analysis, we determined body conductance, a functional measure of heat loss, using open flow respirometry. We used indirect calorimetry to determine oxygen consumption rates (\(\dot{V}{\text{O}}_{2}\)) at ambient temperatures ranging from ~5 to 32 °C. These data were used to determine conductance of both winter (14–23 February; n = 13)- and summer (8 August to 4 September; n = 13)-acclimatized birds following Scholander et al. [30]. In all cases, positive pressure gas flow (room air) was regulated by a mass flow meter (0–5 L/min, Sierra Instruments, Monterey, CA, USA) controlled by a flow controller (model MFC-2; Sable Systems, Las Vegas NV, USA) at 1.4 L/min. Air then flowed into a 2.5-liter water-jacketed Plexiglass chamber containing the animal before passing through a CO2 analyzer (FoxBox Portable Oxygen Analysis System, Sable Systems, Las Vegas NV, USA), a drierite column and an O2 analyzer (FoxBox System). Carbon dioxide was not removed from the gas at any point. Voltage outputs from the flow controller and gas analyzers were collected at a rate of one sample per second using the Expedata program (Sable Systems).
Prior to each measurement, the chamber was brought to a target temperature by passing water through the outer jacket and a temperature-controlled circulating water bath (model F12, Julabo USA, Allentown PA, USA). During measurements, the bird was enclosed in the metabolic chamber, Ta was monitored with a copper/constantan thermocouple thermometer suspended in the center of the chamber, above the animal (HH806AU, Omega Inc., Stamford CT, USA) and the apparatus was covered with a dark box to keep the bird calm. Activity was monitored by listening for movement in the chamber. The bird remained in the chamber for 45–60 min to achieve a steady-state metabolic rate. All respirometry experiments were conducted between late morning and at least 2 h before sunset.
Data for analysis were selected from a 5-min period while the bird was at rest and the gases were at a steady state. \(\dot{V}{\text{O}}_{2}\) was calculated using the equation from the Expedata manual (Sable Systems):
$$\dot{V}{\text{O}}_{2} = {\text{STP}}\cdot{\text{FR}}\cdot\left( {\left( {{\text{F}}_{\text{i}} {\text{O}}_{ 2} {-}{\text{F}}_{\text{e}} {\text{O}}_{ 2} } \right){-}{\text{F}}_{\text{e}} {\text{O}}_{ 2} \times \left( {{\text{F}}_{\text{e}} {\text{CO}}_{ 2} {-}{\text{F}}_{\text{i}} {\text{CO}}_{ 2} } \right)} \right)/\left( { 1{-}{\text{F}}_{\text{e}} {\text{O}}_{ 2} } \right),$$
where STP is standard temperature and pressure corrections, FR flow rate, FiO2 and FeO2 fractional content of oxygen in incurrent and excurrent air, respectively, and FiCO2 and FeCO2 fractional content of carbon dioxide in incurrent and excurrent air, respectively.
Temperature logging
To determine continuous body temperature, after measurement of body composition, birds were anesthetized with inhaled isoflurane (5 % during induction and 1 % for maintenance in oxygen at 1L/min) and a 1-cm incision was made across the abdomen. A Thermochron iButton (model DS1922L; iButtonLink, LLC, Whitewater WI, USA) was implanted into the abdominal cavity and the incision closed with nylon suture and cyanoacrylate glue. The iButton was programed to record body temperature once every 5 min at a resolution of ±0.0625 °C, with an approximately 36-h delay before the first recording was made. After surgery, birds were transported to the Miami University Ecology Research Center (~30 min) and released into an outdoor aviary. Birds were given commercial birdseed and water ad libitum. After 1–2 weeks, the animals were recaptured, euthanized and the data logger was recovered. Only complete days were analyzed. We collected a total of 24 days of Tb data from a total of 4 individual cardinals in summer (24 August to 3 September) and a total of 69 days from 6 cardinals in winter (16 January to 6 March).
Statistics
For body, fat and lean mass, we compared summer and winter animals using a Student’s t test in JMP (version 11). To determine the lower critical temperature for each season, we initially performed a piecewise regression from the respirometry data using PROC NLIN in SAS (ver 9.3; SAS Institute, Cary NC USA). To compare the conductance across season, we then built a model including temperature and season effects in which the individual season inflection points (lower critical temperature) were translated to the same zero point and used as a fixed constant in a subsequent ANCOVA performed within the context of a segmented regression model with both seasons [1] using SAS (ver 9.3). To compare seasonality of body temperature, we compared a model with season, time of day and their interactions as fixed effects and animal and day as random effects to an identical model lacking season and the season by time interaction. A Chi-squared test was then used to determine if season led to a significant difference in the models. We also report the Akaike information criterion (AIC) values and the Bayesian information criterion (BIC) values as further measures of model fit. Analysis was performed using R (v3.1.1; [26]). The level of significance was set at p < 0.05 in all cases. All data are reported as mean ± standard error except the lower critical temperatures which are mean (95 % confidence interval). The number of observations is listed in the corresponding methods above.