Field metabolic rates (FMRs) of 61 species of mammals, as measured wit
h doubly labelled water, range from 29 kJ day-1 (0.34 W) in pipistrel
bats to 49 MJ day-1 (570 W) in northern elephant seals, which is a ran
ge of 1678 times. Most of this variation is due to differences in body
mass; the least-squares, log-log regression of mammalian FMR on body
mass (kJ day-1 = 5.27 g0.723) has a coefficient of determination (r2)
of 0.961, indicating that variation in log(body mass) accounts for 96%
of variation in log(FMR). The scaling of FMR in marsupials (kJ day-1
= 10.83 g0.582, 17 species) differs significantly from that in eutheri
an mammals (kJ day-1 = 4.63 g0.762, 44 species), and the respective r2
-values (0.978 and 0.972) indicate that these taxonomic infraclasses e
xplain another 1% of variation in log(FMR). After adjusting for mass a
nd infraclass effects, residual variation is still substantial (2.5-fo
ld among marsupials and 6-fold among eutherians). What accounts for th
is variation? Neither taxonomic order (or family within the Marsupiali
a), diet category (e.g. herbivore, camivore), nor habitat (e.g. marine
, tundra) explained much residual variation, except that desert-dwelli
ng eutherians had significantly lower FMRs than expected for eutherian
s of their mass. The failure of taxonomic and ecological categories to
account for residual variation may be due, in part, to small sample s
izes and skewed distributions of these categories along the mass axis,
but it seems likely that other sources of variation, such as season,
sex, age, ambient temperature, daily behaviour pattern and food availa
bility may have large effects on FMR that are not accounted for in thi
s analysis.