Building weighted networks for plankton communities from semi-quantitative data

Quantitative weighted network models are valuable approaches widely used for understanding properties of the system, including health and good environmental status of marine ecosystems. Although plankton community data in terms of composition and abundance are widely available additional semi-quantitative ecological information are needed to build weighted networks for plankton communities (D’Alelio et al. 2016a). The goal of this study is the definition of an approach for developing quantitative networks from semi-quantitative data of plankton communities. The approach is based on a controlled interactive process that is used to develop synthetic networks that are then tested for realism using simple and general emerging properties by node and by the whole network as validating tests. We used realistic data of the plankton community that consist of biomass values, realistic ranges of the production rate per unit of biomass as well as rates of mixotrophy (from 0 to 1, with 1 = total phototrophy and 0 = total heterotrophy) specific for all taxa of the plankton community. For each consumer realistic range for consumption per unit of biomass and unassimilated rate were given (D’Alelio et al. 2016b). Furthermore, semi-quantitative indications of the strength of interactions between each taxa were defined (improbable/weak link, likely/ important link, very likely and strong link). An iterative MCMC approach was used by extracting independent and random values for parameters from their range, through a uniform distribution, and assigning random quantitative values of diet proportions (between 0 and 1) constrained by semiquantitative strengths. Synthetic models were tested for their respect of energy conservation principles and physiological realism i.e., maximum gross food conversion efficiency, positive emerging respiration and natural mortality for each living functional group. A large group of valid models (ensemble of 1000 networks) were obtained for each site using the process considering opportunity for some imports to occur or not. Whole network indicators we applied to valid models, e.g., the community primary production and respiration rate (Odum 1969), ascendancy and overhead (Costanza and Mageau 1999). Preliminary results on plankton community networks for a coastal lagoon case study indicate a system, far from its climax, with great potential for further development. The procedure developed could be applied to several marine sites possibly linked to capabilities of the system to provide services.