A new approach to testing the agreement of two phytoplankton quantification techniques: Microscopy and CHEMTAX
Armbrecht, LH, Wright, SW, Petocz, P & Armand, LK 2015, 'A new approach to testing the agreement of two phytoplankton quantification techniques: Microscopy and CHEMTAX', Limnology and Oceanography: Methods, vol. 13, no. 8, pp. 425.
Strengths of numerical relationships between phytoplankton abundance estimates made by microscopy and CHEMTAX have often been tested using regression analysis. To specifically test agreement, where data points lie along a line of equality, the Bland and Altman technique is commonly used in the medical literature and applied here to phytoplankton analysis for the first time (simultaneously with regression). Our analyses are based on a sample set collected off Coffs Harbour (âˆ¼ 30Â°S), Eastern Australia. While comparing abundance estimates of different phytoplankton pigment-types derived from microscopy and CHEMTAX we specifically aim at: (1) determining the usefulness of the Bland and Altman technique in comparing phytoplankton abundance estimates made by both quantification techniques, and (2) identifying the lowest taxonomic level (i.e., interclass or intraclass level) at which phytoplankton abundance estimates agree. Our results suggest that Bland and Altman analysis is highly suited to compare phytoplankton abundance estimates made by microscopy and CHEMTAX. It delivered a quantifiable difference of chlorophyll a concentrations between abundance estimates, which may benefit the future calibration of phytoplankton quantification techniques. Both Bland and Altman and regression analyses were suited to resolve imbalances between phytoplankton abundance estimates made by microscopy and CHEMTAX. Best agreement was found within dinoflagellates (class-level), poor agreement within three diatom pigment-types (intraclass level). We attributed the poor agreement within diatom abundance estimates to classification errors of microscopically determined taxa into pigment-types. Intraspecific pigment composition seems more variable than generally assumed, calling for studies resolving this variation to further conform abundance estimates made by microscopy and CHEMTAX.