Distribution of twelve moist forest canopy tree species in Liberia and Cote d'Ivoire: response curves to a climatic gradient

The occurrence and abundance of 12 canopy tree species from the moist tropi cal forests of West Africa have been studied in relation to a climatic grad ient. We focused on environmental factors related to water availability: an nual amount of rainfall, the length of the dry season, and cumulative water deficit. Species occurrence and abundance data are used for 39 forest site s in Liberia and southwest Cote d'Ivoire. Species responses are modelled us ing a set of five increasingly complex models, ranging from a no-trend mode l to a skewed bell-shaped response curve. The study species show different distribution patterns. Most of them sugges t a close relationship to climatic conditions. Fitting of species occurrenc e data to each of the three climatic factors results in most cases in simpl e models. In only one out of 36 cases a bell-shaped response curve is neede d to describe the data. Four of the 12 species show no response to the clim atic factors when only occurrence is evaluated. When abundance data are used, in 33 of the 36 cases significant response mo dels are found. In general these are much more complex than in the cases of species occurrence data: in 10 of the 36 cases a bell-shaped response mode l is found to describe the data best. This is in contrast with the widespre ad belief that species response curves are bell-shaped: within the forest z one in the area studied this is not generally the case. The importance of the three climatic factors for the distribution of the sp ecies is evaluated: for four species mean annual rainfall is the most impor tant variable, for four species the length of the dry period, and for one s pecies cumulative water deficit. Consequently, the assumption that mean ann ual rainfall is the most important factor determining tree species distribu tion in West African forests is not correct. Species response models to climatic factors show where species have their g eographical optima. Implications for forest management are briefly discusse d.