Determining the probability of cyanobacterial blooms: The application of Bayesian networks in multiple lake systems

A. Rigosi, P.C. Hanson, D.P. Hamilton, Matthew Hipsey, J.A. Rusak, J. Bois, K. Sparber, I. Chorus, A.J. Watkinson, B. Qin, B. Kim, J.D. Brookes

Research output: Contribution to journalArticlepeer-review

117 Citations (Scopus)


© 2015 by the Ecological Society of America A Bayesian network model was developed to assess the combined influence of nutrient conditions and climate on the occurrence of cyanobacterial blooms within lakes of diverse hydrology and nutrient supply. Physicochemical, biological, and meteorological observations were collated from 20 lakes located at different latitudes and characterized by a range of sizes and trophic states. Using these data, we built a Bayesian network to (1) analyze the sensitivity of cyanobacterial bloom development to different environmental factors and (2) determine the probability that cyanobacterial blooms would occur. Blooms were classified in three categories of hazard (low, moderate, and high) based on cell abundances. The most important factors determining cyanobacterial bloom occurrence were water temperature, nutrient availability, and the ratio of mixing depth to euphotic depth. The probability of cyanobacterial blooms was evaluated under different combinations of total phosphorus and water temperature. The Bayesian network was then applied to quantify the probability of blooms under a future climate warming scenario. The probability of the "high hazardous" category of cyanobacterial blooms increased 5% in response to either an increase in water temperature of 0.8°C (initial water temperature above 24°C) or an increase in total phosphorus from 0.01 mg/L to 0.02 mg/L. Mesotrophic lakes were particularly vulnerable to warming. Reducing nutrient concentrations counteracts the increased cyanobacterial risk associated with higher temperatures.
Original languageEnglish
Pages (from-to)186-199
JournalEcological Applications
Issue number1
Publication statusPublished - 2015


Dive into the research topics of 'Determining the probability of cyanobacterial blooms: The application of Bayesian networks in multiple lake systems'. Together they form a unique fingerprint.

Cite this