Nutrients were added to Peter (2013-2015, 2019) and Tuesday (2013-2015)
lakes to cause algal blooms. Details on nutrient additions (start/end dates,
loading rates, N:P ratios) are described in Buelo et al. 2022 (Ecological
Applications, link below), Wilkinson et al. 2018 (Ecological Monographs 88:
188-203), and Pace et al. 2017 (Proceedings of the National Academy of
Sciences USA 114: 352-357). These publications including supplements should
be consulted for details. These lakes have been used for whole-ecosystem
experiments over the past decades; see Carpenter and Pace 2018 (Limnology
and Oceanography Letters 3(6): 419-427) for an overview.
For chlorophyll-a concentrations, water samples were collected from 0.5m
depth daily at the center of each lake. Water samples were filtered in the
lab, frozen for at least 24 hours, extracted in methanol, and measured on a
fluorometer following methods described in the publications above.
Phycocyanin fluorescence, dissolved oxygen saturation, and pH were
measured from a raft at the center of the lake every 5 minutes by automated
sensors at a depth of 0.75m. Sensors were calibrated monthly according to
manufacturer recommendations. Phycocyanin values are from Hydrolab DS5X data
sondes using manufacturer-provided calibration curves and should be
interpreted as relative and not exact, physical units. Dissolved oxygen and
pH were measured by YSI 6600, Hydrolab DS5X, and YSI EXO3 data sondes. High
frequency data were gap filled as described in Wilkinson et al. 2018
(Ecological Monographs 88: 188-203) and averaged to daily
measurements.
An R package utilizing these data for the evaluation of early warning
statistic methods to predict algal blooms is available on GitHub
(https://github.com/cbuelo/tvsews) and archived on Zenodo
(https://doi.org/10.5281/zenodo.5874868).