Some species flourish when many do not: a pattern in data on ecological communities

Patterns in species × sample tables of communities depend above all on the organisms of the data sets and the conditions involved. Patterns that surpass individual sets are of special interest. Our question, looking for a shared pattern in 12 sets, is if relative abundances among species are independent of the sample, or formulated alternatively, if species have abundances that are correlated with total abundances over samples. For exploration we study the overdispersion/aggregation of the data. A relatively high variation in the total abundances of samples is noticed, indicating an effect of environmental variation. Overdispersion imposes constraints on the accommodation of relatively high abundance values to samples with a relatively low total abundance. The null hypothesis of ‘no association’ is modelled by permutation/resampling of the data at the level of the individual. A correlation study of actual and permuted sets is performed. All actual sets contain a significant number of species that defy our question. These species flourish when many do not. The relation of our question with issues in theoretical ecology, such as the assumption of a neutral effect of environmental conditions and/or of neutral characteristics of species, is discussed.