Our earlier tests of the efficiency of our system indicated that although we underestimated actual oxygen consumption of the ethanol lamp, we did so with a fair degree of precision; probably because flow rates were closely controlled. During our metabolic measurements, chamber flow rates also were closely controlled at 3.0 L/min, and we believe, therefore, that these measurements also were carried out with a high degree of precision. Consequently, all measured values of oxygen consumption and water production were considered to be 84% of their actual value and were adjusted to 100% before being included in this report.
The calibration of all temperature-sensitive radio transmitters drifted over time. Transmitters were calibrated before they were surgically implanted and again after they were removed from the animals. Although the drift of each transmitter was unique, it was also linear (S. Tomkiewicz, Telonics, Inc., pers. com.). All body temperature measurements were corrected from timed extrapolations of the difference between starting and ending calibrations.
Values of oxygen consumption, evaporative water loss, and body temperature were plotted as a function of chamber air temperature. Linear regressions of oxygen consumption at temperatures below the thermoneutral zone (T_{ n}), and evaporative water loss at temperatures above freezing, were determined with the SAS (1982) GLM procedure. Lower critical temperature (T_{ lc}) was determined graphically from intersection of the line representing [.H]_{ b} and the regression line representing oxygen consumption below T_{ n}. Slopes and intercepts of regression lines, as well as other mean values, were compared with _t_-tests (Statistical Analysis System, 1982; Ott, 1984:138-175). Unless indicated otherwise, data are expressed as mean ± standard deviation (s.d.).
ESTIMATING INTRINSIC RATE OF NATURAL INCREASE
We employed the method first described by Cole (1954) to calculate r_{ max}:
1 = e^{ -r_{ max}} + b·e^{ -r_{ max}(a)} - b·e^{ r_{ max}(n+1)} Eq. 2
where a is potential age of females first producing young, b is potential annual birth rate of female young, and n is potential age of females producing their final young. After life-history data were substituted into Eq. 2, r_{ max} was determined by trial and error substitution (Hennemann, 1983).