Recent evidence suggests that avian facultative hypothermic responses are more common than previously thought. Traditionally, several categories of avian hypothermic responses have been recognized, and are frequently differentiated on the basis of minimum body temperature (T[b]) The available data suggest that the capacity for shallow hypothermia (rest-phase hypothermia) occurs throughout the avian phylogeny, but that the capacity for pronounced hypothermia (torpor) is restricted to certain taxa. However, there are currently too few data to test hypotheses concerning the evolution of avian hypothermic responses. Facultative hypothermia occurs over most of the avian body mass (M[b]) range, but is most common in small species. Minimum body temperature during hypothermia (T[min]) is continuously distributed from 4.3 °C to ca. 38°C. The continuous T[min] distribution, as well as recent evidence that the T[b] ranges of different avian physiological states may overlap, question the biological reality of specific T[b] limits. Pattens of thermoregulation during avian hypothermic responses are relatively variable, and do not necessarily follow the entry-maintenance-arousal patterns that characterize mammalian responses. Avian hypothermic responses are determined by a suite of ecological and physiological determinants. I investigated normothermic thermoregulation and hypothermic responses to restricted food in the speckled mousebird Colius striatus in the context of the distinction between normothermia, rest-phase hypothermia, and torpor. The lowest T[b] recorded in a bird which was able to arouse spontaneously was 18.2°C. However, I was unable to clearly discern between normothermic, hypothermic and torpor T[b] ranges. Furthermore, hypothermic responses did not accord with the patterns typically observed in birds and mammals. Metabolic suppression normally associated with entry into torpor and the defence of a torpor T[b] setpoint was largely absent. Laboratory data for C. striatus, as well as published data for Colius colius suggest that clustering behavior plays an important thermoregulatory role in mousebirds. Hence, I investigated thermoregulation under semi-natural conditions in C. striatus. In particular, I was interested in the interaction between clustering behavior and hypothermic responses during energy stress (restricted feeding). In contrast to clustering birds, rest-phase thermoregulation in single birds was characterised by linear decreases in T[b] and the birds did not appear to defend a specific T[b] setpoint. During restricted feeding, both clustering and single birds exhibited significant decreases in rest-phase T[b]. The extent of these facultative hypothermic responses was greater in single birds than in clustering birds, supporting the prediction that clustering behavior moderates the use of facultative hypothermia. I also tested the prediction that in free-ranging C. colius, the use of heterothermy should be rare, even at the coldest time of the year. I recorded mid-winter rest-phase body temperatures (T[b]) in a flock of free-ranging C. colius in an arid habitat in the Karoo, South Africa. The mousebirds' rest-phase T[b] was fairly labile, but was maintained above 33°C, despite T[a]s as low as -3.4 °C. The mousebirds showed no evidence of torpor under natural conditions; a facultative hypothermic response, during which T[b] was reduced to 29 - 33°C, was only observed on one occasion. The observed patterns of thermoregulation supported my predictions, and suggest that thermoregulation in clustering C. colius in the wild is significantly different to that of single birds under laboratory conditions. My results also suggest that the pronounced capacity for heterothermy usually associated with mousebirds is not necessarily representative of their patterns of thermoregulation under natural conditions. The capacity for avian torpor appears to be dependent on phylogeny. To investigate phylogenetic constraints on the capacity for torpor, I measured metabolic responses to food deprivation in a small, arid-zone passerine, the red-headed finch (Amadina erythrocephala). I observed significant reductions in rest-phase energy expenditure and body temperature (T[b]) in response to restricted feeding. The maximum extent of T[b] suppression (ca. 5°C) and energy savings (ca. 10%) were consistent with those reported for a number of other passerines. The lowest T[b] I observed in a bird able to arouse spontaneously was 34.8°C. My data support the hypothesis that the capacity for heterothermy in passerines is phylogenetically constrained, and that the majority cannot employ torpor in response to energetic stress. Selection for the capacity for torpor is presumably similar to the selection pressures acting on other avian energetic traits, such as basal metabolic rate (BMR). I tested the generality of a recent model linking the slow-fast mammalian metabolic continuum to global patterns of climatic predictability using BMR data for 219 non-migratory bird species. Avian BMR varied significantly between zoogeographical zones, with Afrotropical, Indomalayan and Australasian species generally exhibiting lower BMR than Holarctic species. In addition, the magnitude of differences between arid and mesic species varied between zones. In the Nearctic, these differences were pronounced, whereas no significant differences were evident for Afrotropical or Australasian species. A slow-fast metabolic continuum similar to that described in mammals appears to exist for birds, with higher BMR associated with predictable, seasonal environments and lower BMR with less predictable environments, in particular those affected by the El Niño Southern Oscillation. I constructed a generalised, conceptual model which attempts to predict the occurrence of torpor using phylogeny, M[b] constraints, a trade-off between energetic benefits and potential ecological costs, and specific ecological factors. A recent hypothesis suggests that endotherm heterothermy is monophyletic, and predicts that torpor should be more widespread in phylogenetically older taxa. Once phylogeny is considered, the most important determinant of avian torpor is M[b]. I used an existing model of endotherm torpor to predict the relationship between M[b] and minimum T[b] during torpor. The available data show that the lower limit of torpor T[b] is determined by the M[b]-dependent costs of rewarming following a torpor bout. Finally, I constructed a model based on the assumption that torpor is adaptive if the energetic benefits exceed the potential ecological costs. The model predicted that torpor should be more prevalent in species near the extremes of the avian metabolic continuum. The available data provide tentative support for this prediction. In addition to generalised factors such as phylogeny and M[b], specific aspects of a particular species' ecology need to be considered when predicting the occurrence of avian torpor. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 2001.