Habitat niche community-level analysis of an amphibian assemblage at Lake Nabugabo, Uganda

Abstract. Community structure was studied across six different habitat types in an amphibian assemblage constituted by 24 species belonging to five families, from Lake Nabugabo, Uganda. We employed a suite of different statistical methods, including univariate, multivariate, and Monte Carlo procedures to investigate the randomness/nonrandomness and the seasonal effects (wet versus dry season) of the community assembly. We calculated for each species in each habitat type an index of relative abundance by using a time constrained counting technique, with 48 1-h counts for each habitat type. Co-occurrence was analysed by C score with 30 000 simulations; resource partitioning patterns by RA2 and RA3 algorithms with 30 000 simulations; and apparent dissimilarity among species in terms of habitat use by UPGMA dendrograms. After pooling data from wet and dry seasons, it resulted that the amphibian community was non-randomly assembled according to C-score analyses, but both RA2 and RA3 were unable to uncover any competitive structure for the dataset. Seasonal effects were evident, and although C score analyses confirmed a nonrandom structure for the community under study (particularly in wet season), RA3 showed that species with high relative abundance tended to significantly concentrate in one habitat type (swamp forest) rather than to partition the habitat resource. UPGMA dendrograms grouped the species differently in dry versus wet seasons. Overall, the comparative evidence of 1) non-random community structure according to C-score analysis, and 2) absence of resource partitioning according to niche overlap null models analysis, suggests that community organization in Lake Nabugabo amphibians is generated by habitat affinities rather than by interspecific competition.