Aim We used fossil and phylogenetic evidence to reconstruct climatic niche evolution in Nothofagus, a Gondwana genus distributed in tropical and temperate latitudes. To assess whether the modern distribution of the genus can be explained by the tropical conservatism hypothesis, we tested three predictions: (1) species from all Nothofagus subgenera coexisted under mesothermal climates during the early Eocene; (2) tolerance to microthermal climates evolved during the Eocene-Oligocene cooling from an ancestor that grew under mesothermal conditions; and (3) the climatic niche in Nothofagus is phylogenetically conserved.Location Australia, New Zealand, New Caledonia, Papua-New Guinea and South America.Methods We estimated the palaeoclimate of the Early Eocene, fossil-bearing Ligorio Marquez Formation (LMF, Chile), using coexistence and leaf physiognomic analysis. We reconstructed ancestral climatic niches of Nothofagus using extant species distributions and a time-calibrated phylogeny. Finally, we used the morphological disparity index and phylogenetic generalized least squares to assess whether climatic variables follow a Brownian motion (BM) or an Ornstein-Uhlenbeck (OU) model of evolution.Results Our palaeoclimatic estimates suggest mesothermal conditions for the LMF, where macrofossils associated with subgenera Lophozonia and possibly Fuscospora, and fossil pollen of Brassospora and Fuscospora/Nothofagus were recorded. These results are not supported by our phylogenetic analysis, which instead suggests that the ancestor of Nothofagus lived under microthermal to marginally mesothermal conditions, with tolerance to mesothermal conditions evolving only in the subgenus Brassospora. Precipitation and temperature dimensions of the realized climatic niche fit with a gradual BM or constrained OU model of evolution. Main ConclusionsOur results suggest that the use of phylogenetic reconstruction methods based only on present distributions of extant taxa to infer ancestral climatic niches is likely to lead to erroneous results when climatic requirements of ancestors differ from their extant descendants, or when much extinction has occurred.

Broad‐scale richness gradients are closely associated with temperature and water availability. However, historical and evolutionary processes have also contributed to shape current diversity patterns. In this paper we focus on the potential influences of Pleistocene glaciation and phylogenetic niche conservatism (the tendency for traits to be maintained during diversification) on the tree diversity gradient in Chile, and we quantify its primary climatic correlates. Tree species richness is greatest at mid latitudes, particularly in the Andes and Coastal ranges, and decreases abruptly to the south and north. Regression tree analysis identified annual precipitation and annual temperature as the primary probable drivers of this gradient. Ice cover during the Last Glacial Maximum was also identified as an ‘important’ variable, but the contemporary and historical predictors are strongly collinear. Geographically weighted regression indicated that the relationships between richness and environmental variables vary regionally: the relationship between tree richness and precipitation is stronger in north‐central Chile, whereas tree richness and temperature are most strongly associated in south‐central Chile. By assigning each species the age of the family to which it belongs and averaging all species in each geographical unit, we also found that species from the oldest families are distributed mainly in mid to high latitudes and species from younger families are distributed mainly at lower latitudes. This pattern is closely associated with annual precipitation. Thus, the ecological component of tree richness follows contemporary climatic gradients of both energy and water, but the aridification of the Atacama Desert was an important driver over evolutionary time. The influence of recent Pleistocene glaciation remains unresolved but it cannot be discounted.