Marine Food Webs: an Exploration of Past Structure and Function to Prepare for the Future

Food webs depict species and their trophic interactions arranged in a network. Changes in food web structure ultimately beget changes in functioning. In recent years, environmental and anthropogenic pressures have reshaped marine communities beyond recognition, with either gains (e.g. through invasions) or losses (e.g. through local extinctions) in biodiversity. Still, we do not know how it has impacted the food web. Assessing changes in the structure and function of food webs is complex as the components at the core of food webs (the composition of species and the architecture of trophic interactions) can vary in intricate ways. As a result, sampling temporally resolved food webs is often difficult and, consequently, few studies have investigated temporal changes in structure and function. The few studies available have often relied on data with low temporal resolution (such as, a before-and-after state). To study temporal variation in food web structure, we need to improve how we build networks, and develop a framework that allows us to identify diverse changes in food webs over time.
The overall aim of this thesis is two-fold. First, the thesis explores temporal changes in food web structure and function, and aims at disentangling changes that originate from shifts in species composition, dominance of species (abundance, biomass), trophic links (caused by changes in composition), as well as energy fluxes. In Paper I and II, I constructed time series of food web metrics for two local food webs (the German Bight in the North Sea and the Gulf of Riga in the Baltic Sea), including metrics developed to identify topological changes occurring without the complete loss of species. I built yearly snapshots of food webs by subsampling taxa and their trophic links from metawebs (a food web topology that contains all possible species and trophic interactions) using time series of species abundances or biomasses. The findings highlight that temporal changes in food webs are complex and varied: study of food web temporal development necessitates the use of complementary metrics that integrate and expose the various underlying causes of change (e.g. species composition, dominance, trophic interactions, energy fluxes). For instance, variation in food web structure resulting from changes in species composition may further impact the functioning of the food web, and simultaneous assessment of both types of changes is possible using a set of complementary qualitative and quantitative food web metrics.
Second, I aim to find new ways to improve the building, use, and applicability of food webs. Being able to build realistic food webs is essential if we want to study their structure and functioning, or integrate food webs into monitoring and management of ecosystems. Future advances require tackling fundamental challenges: how to aggregate species and their links when diet data is scarce; or which trophic links to include, for instance, when species composition changes over time. In Paper III, I tested the sensitivity of food web structural metrics to several aggregation procedures. I show that metrics differ in their sensitivity to aggregation, and that not all aggregation methodologies perform best at maintaining food web structure. In Paper IV, I investigated trait-associations through which consumers and resources interact at the local scale. Functional traits of species govern where species live and with whom they can interact. I found that food webs structure along a continuum determined by similarities in habitat characteristics, where species in similar habitats most likely share traits. By combining food webs and traits, I identified trait profiles that portray interacting consumers and resources along the basic vertical organization of food webs (trophic levels). Trait profiles characterise interacting consumers and resources, and can thereby help with inference of trophic interactions for the purpose of building food webs.
Keywords: food web, coastal community, environmental changes, time series, food web aggregation, functional traits, trait matching, bipartite network.

Turku, December 24th, 2021
Pierre Olivier