Competition is a major regulatory factor in population and community dynamics. Its effects can be either direct in interference competition or indirect in exploitative competition. The impact of exploitative competition on population dynamics has been extensively studied from empirical and theoretical points of view, but the consequences of interference competition remain poorly understood. Here we study the effect of different levels of intraspecific interference competition on the dynamics of a size-structured population. We study a physiologically structured population model accounting for direct individual interactions, allowing for a gradient from exploitative competition to interference competition. We parameterize our model with data on experimental populations of the collembolan Folsomia candida. Our model predicts contrasting dynamics, depending on the level of interference competition. With low interference, our model predicts juvenile-driven generation cycles, but interference competition tends to dampen these cycles. With intermediate interference, giant individuals emerge and start dominating the population. Finally, strong interference competition causes a novel kind of adult-driven generation cycles referred to as interference-induced cycles. Our results shed new light on the interpretation of the size-structured dynamics of natural and experimental populations.

1. Because of recent technological improvements in the way computer and digital camera perform, the potential use of imaging for contributing to the study of communities, populations or individuals in laboratory microcosms has risen enormously. However its limited use is due to difficulties in the automation of image analysis. 2. We present an accurate and flexible method of image analysis for detecting, counting and measuring moving particles on a fixed but heterogeneous substrate. This method has been specifically designed to follow individuals, or entire populations, in experimental laboratory microcosms. It can be used in other applications. 3. The method consists in comparing multiple pictures of the same experimental microcosm in order to generate an image of the fixed background. This background is then used to extract, measure and count the moving organisms, leaving out the fixed background and the motionless or dead individuals. 4. We provide different examples (springtails, ants, nematodes, daphnia) to show that this non intrusive method is efficient at detecting organisms under a wide variety of conditions even on faintly contrasted and heterogeneous substrates. 5. The repeatability and reliability of this method has been assessed using experimental populations of the Collembola Folsomia candida. 6. We present an ImageJ plugin to automate the analysis of digital pictures of laboratory microcosms. The plugin automates the successive steps of the analysis and recursively analyses multiple sets of images, rapidly producing measurements from a large number of replicated microcosms.