The investigation of changes in abundance of calcareous nannofossil associations is fundamental for palaeoclimatic and palaeoceanographic reconstructions. In general, since the number of recognised species is high, the numerical and graphical treatments of the data play an important role to validate hypotheses about the influence of physico–chemical factors on the quality of past environments, as reflected by changes in abundance. Moreover, the application of multivariate procedures may be useful for improving interpretation of the overall data structure. Calcareous nannofossil associations analysed in this work are related to core GNS84-C106 collected in the Gulf of Salerno (Campania, southern Italy) under a national interdisciplinary research programme aimed at recognising the impact of natural events on biological communities during the last 20 kyr. Biplot analysis of compositional data was applied to indicate the presence of subcompositions or, in other words, to identify strong relationships among a limited number of species that can be interpreted from a palaeoenvironmental point of view. This investigation was followed by the application of simplicial principal component analysis with the aim of determining, from a numerical point of view, the laws for describing and modelling the behaviour of the species involved in these subcompositions. In this work, all statistical methods were applied by taking rigorously into account the sample space of compositional data (i.e. data matrices whose rows sum to 100%). From a general point of view, the results obtained, related to the detection of new parameters to be used for monitoring the quality of the past environment, have encouraged us to promote the application of statistical methodologies that take into account the correct sample space for compositional data. In this context, inferential applications and modelling can be developed under a correct theoretical framework.