PT  - JOURNAL ARTICLE
AU  - Emanuele Giacomuzzo
AU  - Ferenc Jordán
TI  - Food web aggregation: effects on key positions
AID  - 10.1101/2021.04.18.440319
DP  - 2021 Jan 01
TA  - bioRxiv
PG  - 2021.04.18.440319
4099  - http://biorxiv.org/content/early/2021/04/22/2021.04.18.440319.short
4100  - http://biorxiv.org/content/early/2021/04/22/2021.04.18.440319.full
AB  - Providing standard definitions of what should be considered as a node in food webs is still an unsolved problem. Especially for comparative and predictive food web modelling, a more systematic understanding is needed for the effects of trophic aggregation procedures. Aggregation is unavoidable during data management. Therefore, it is crucial to know whether food web properties are conserved during this process.Here, we study how different aggregation methods change the positional importance of species in food webs. In particular, we investigated the effects of various aggregation algorithms on 24 indices of importance. Our work was carried out on 76 aquatic food webs coming from the Ecopath with Ecosim database (EcoBase). We considered six main types of aggregation, according to the way that the nodes were clustered. These were (i) hierarchical clustering based on the Jaccard index, (ii) hierarchical clustering based on the regular equivalence index (REGE), (iii) maximisation of directed modularity, (iv) maximisation of modularity according to modules in which species fed on the same preys, (v) maximisation of modularity according to modules in which species are fed upon by the same predators, and (vi) clustering through the group model.Hierarchical clustering based on the Jaccard index and REGE index outperformed the other four methods on maintaining the relative importance of species for all the indices of importance (except for the contrastatus index (s′) and betweenness centrality (BC)). The choice between these two methods should follow our research question and the importance index we are interested in studying. The other four aggregation methods change more the centrality of species, especially the one based on maximising directed modularity. When using these aggregation algorithms, one has to keep in mind that the network will not only be smaller but also provides different information.Competing Interest StatementThe authors have declared no competing interest.