Wed, Aug 04, 2021:On Demand
Background/Question/Methods
Ecosystem functions arise from the activities and interactions of individual organisms, but the link between community composition and ecosystem functioning is difficult to quantify particularly in natural systems. We demonstrate a new approach for examining this link: fitting a dynamic model to a collection of local time series data sets and relating the variation in model parameters describing ecosystem flows to the simultaneously observed variation in community composition. We fitted a dynamic nutrient-phytoplankton-zooplankton (NPZ) model to a series of 152 bioassay experiments conducted at ca. biweekly intervals throughout three growing seasons at six different sites along the Finnish coast of the Baltic Sea. Spatiotemporal variation in NPZ model parameters (phytoplankton maximum growth rate, phytoplankton nutrient uptake affinity, zooplankton grazing rate and phytoplankton C:N and ChlA:N ratios) was quantified by combining the dynamic core model with a hierarchical Bayesian data model that describes the relationship between the different experiments. In a post-hoc analysis utilizing community composition data from the experiments, published compilations of binary and continuous phytoplankton traits and phylogenetic imputation of missing trait values, we compared the patterns in NPZ model parameters to patterns in phytoplankton functional composition.
Results/Conclusions The NPZ model with locally varying parameters performed well in reproducing the growth dynamics observed in the experiments. There was significant seasonal and spatial variation in the estimated NPZ model parameters, but this variation was only weakly correlated with variation in the binary trait composition of the phytoplankton community. Moreover, there was considerable mismatch between the system-level phytoplankton maximum growth rate and nutrient uptake affinity estimated from the experimental time series and weighted averages of taxon-specific values of these same traits as reported in published compilations based on monoculture measurements. Only phytoplankton C:N ratio showed some accordance between the system-level estimates and weighted averages of taxon-specific values. We conclude that 1) hierarchical dynamic modelling is a useful tool for analyzing variation in ecosystem functioning, 2) the functioning of natural plankton systems shows variation that is related to phytoplankton functional composition, but it is difficult to predict the dynamics of a given system from the traits of the taxa that make out the community.
Results/Conclusions The NPZ model with locally varying parameters performed well in reproducing the growth dynamics observed in the experiments. There was significant seasonal and spatial variation in the estimated NPZ model parameters, but this variation was only weakly correlated with variation in the binary trait composition of the phytoplankton community. Moreover, there was considerable mismatch between the system-level phytoplankton maximum growth rate and nutrient uptake affinity estimated from the experimental time series and weighted averages of taxon-specific values of these same traits as reported in published compilations based on monoculture measurements. Only phytoplankton C:N ratio showed some accordance between the system-level estimates and weighted averages of taxon-specific values. We conclude that 1) hierarchical dynamic modelling is a useful tool for analyzing variation in ecosystem functioning, 2) the functioning of natural plankton systems shows variation that is related to phytoplankton functional composition, but it is difficult to predict the dynamics of a given system from the traits of the taxa that make out the community.