Many animals can modify the environments in which they live, thereby changing the selection pressures they experience. A common example of such niche-construction is the use, creation, or modification of environmental resources for use as nests or shelters. Because these resources often have correlated structural elements, it can be difficult to disentangle the relative contribution of these elements to resource choice, and the preference functions underlying niche-construction behaviour remain hidden. Here, we present an experimental paradigm that uses 3D-scanning, -modelling, and -printing to create replicas of structures that differ with respect to key structural attributes. We show that a niche-constructing, shell-dwelling cichlid fish, Neolamprologus multifasciatus, has strong open-ended preference functions for exaggerated shell replicas. Fish preferred shells that were fully intact and either enlarged, lengthened, or had widened apertures. Shell intactness was the most important structural attribute, followed by shell length, then aperture width. We disentangle the relative roles of different shell attributes, which are tightly correlated in the wild, but nevertheless differentially influence shelter choice and therefore niche construction in this species. We highlight the broad utility of our approach when compared to more traditional methods (e.g. two-choice tasks) for studying animal decision-making in a range of contexts.

This dataset was collected according to the methods detailed in Bose et al. "Structural manipulations of a shelter resource reveal underlying preference functions in a shell-dwelling cichlid fish". In brief, Neothauma tanganyicense snail shells were micro-CT scanned, modelled, and 3D printed to create shell replicas. In each replica, we experimentally manipulated one structural attribute, either enlarging it or diminishing it, independently of other structural attributes. We then presented these shell replicas to Neolamprologus multifasciatus individuals in a series of shell choice tasks. We analyzed the data using a preference function approach. See Methods section and Supplementary Materials for more details.

There are two data files.

1. Shell_preferences.csv contains all the behavioural results from the choice tasks wherein shell replicas were presented to individual fish. First several columns give information on the individual fish participants, the next several columns given summary information about the individual fish's preference functions (i.e. preference function traits), and the last few columns given information on how many trials the fish participated in. Note that this dataframe also includes the simulated trials needed to generate 'random' preference function traits.

2. Shell_choice_chirality.csv contains the behavioural results from the shell chirality choice tasks. The first few columns given information on the individual fish participants, while the final columns give their actual shell choices.