phytoclassShiny: a Guided Workflow for Phytoplankton Pigment Chemotaxonomy

Liam Bodley Quinlan , Johannes J. Viljoen, Alexander Hayward, Simon W. Wright, Tylar Murray, Sebastian Di Geronimo, Karen J. Westwood, Mohammad Aasim Khan, Susanne Fietz

Estimating phytoplankton community structure using diagnostic pigment biomarkers relies on chemotaxonomic tools like CHEMTAX and the mathematical improvements of the phytoclass R package. However, phytoclass requires programming experience that limits its accessibility to non-coders. We present phytoclassShiny, a free, open-source, offline R Shiny application that guides users through a linear, seven-step workflow - from high-performance liquid chromatography (HPLC) data import and quality control to simulated annealing optimisation and export - without requiring programming skills. Every session’s parameters, data-handling decisions, and results are recorded in a shareable configuration file and audit log, addressing current reporting gaps and supporting full reproducibility. We validated phytoclassShiny's phytoplankton community estimates against the native phytoclass package using a benchmark Southern Ocean dataset. Estimates agreed closely across seven of eight phytoplankton groups (R² ≥ 0.97), with one group showing marginally lower agreement (Pelagophytes, R² = 0.90). This minor variation was attributed to a disclosed difference in benchmark parameters rather than by any alteration of phytoclass's underlying mathematics. Repeated tests using a fixed random seed reproduced results exactly. phytoclassShiny therefore removes the programming burden of phytoclass analysis without any cost in accuracy. However, like all chemotaxonomic methods, successful analysis still relies on the user's foundational taxonomic and ecological knowledge of their study system. phytoclassShiny broadens access to pigment-based chemotaxonomy for researchers without programming experience, and, when used deliberately, can help newcomers build the very foundational experience that emergent phytoclass method requires.