UC San Diego

Phytoplankton play important roles in marine food webs and biogeochemical

cycles. The ability to understand natural changes in phytoplankton populations are a

central goal in biological oceanography; however, the dynamics of phytoplankton

populations are often noisy and difficult to predict. The portfolio effect in ecology is a

concept that relies on the aggregation of data to reduce the variance of individual

populations. We test whether the aggregation of data across multiple species can help

improve the predictions of phytoplankton assemblages one month into the future. Using

empirical dynamic modelling, we assess the predictability of phytoplankton

assemblages in varying group sizes, and show that certain assemblages of

phytoplankton species are more predictable in groups rather than as individual species.