Quantifying the uncertainties in multi-wavelength PAM fluorometry due to innate and irradiance-induced variability of fluorescence spectra

Multi-wavelength Chl a fluorometers are increasingly applied to assess phytoplankton photosynthetic capacity and composition, but their usefulness is limited by uncertainties in fluorescence excitation spectra (FES). We investigated this issue using the Phyto-PAM fluorometer to evaluate the effects of innate and irradiance-dependent variations in background (F) and variable (F-v) FES on analysis of three pigment groups (cyanobacteria, chlorophytes and chromophytes). The effects on group-specific estimates of minimum fluorescence (F-0), a proxy for biomass, and F-v/F-m, the quantum yield of photochemistry, presented some challenges to the interpretation of group-specific results. F-0 estimates usually had a 5-15% margin of error, even when measuring highly uneven mixtures, and applying imperfectly matched calibration FES or stressing samples with photosynthetically active and ultraviolet radiation; errors in F-v/F-m were commonly < 15%. Despite such relatively good accuracy, estimates for F-0 and, especially, F-v/F-m are unreliable for groups at low relative abundance, and results can sometimes be reported for groups not actually present. We report margins of error for different levels of relative abundance to inform interpretation of measurements from natural communities and show that F and F-v spectra for some taxa can differ in ways that produce severe errors in F-0 and F-v/F-m estimates if used uncritically.

Automated summary

Multi-wavelength Chl a fluorometers are increasingly used to assess phytoplankton photosynthetic capacity and composition, but their usefulness is limited by uncertainties in fluorescence excitation spectra (FES). The effects of innate and irradiance-dependent variations in background and variable spectra on the analysis of three pigment groups present challenges to the interpretation of results. Despite relatively good accuracy, estimates for F-0 and F-v/F-m are unreliable for groups at low relative abundance and can sometimes produce errors in results.