Assessment of stream ecosystem health based on oxygen metabolism: Which sensor to use?

Here, we compared sensor precision and drift among four commonly used dissolved oxygen (DO) sensor types under laboratory conditions, and evaluated their accuracy of measurement of stream ecosystem metabolism (EM) and subsequent assessment of stream ecosystem health in the field. Of all tested sensor types, optical sensors had the highest precision, followed by miniaturized polarographic sensors. Traditional polarographic and pulsed polarographic sensors had a lower precision. The traditional polarographic sensors were the only ones exhibiting detectable sensor drift that had to be corrected for, prior to calculations of EM. In parallel EM measurements in two rural and two pristine tropical streams, the four compared DO sensor types estimated relatively similar gross primary production (GPP), ecosystem respiration (R) and GPP:R ratios in each stream, and classified all streams as highly heterotrophic. Nonetheless, parallel GPP and R estimates with different sensors had relative standard deviations of the mean between 6.2% and 34.9%, and between 7.9% and 16.3%, respectively. Thus, sensor accuracy can be a major source of variability in EM estimates. There were only minor differences in stream ecosystem health assessments among different DO sensors, compared to differences between different assessment methods. While all tested sensor types yielded useful ecosystem health classifications, the optical sensor type was the most accurate, and thus, the most suitable for functional impact assessment and monitoring.