MEPS 613:49-66 (2019)  -  DOI: https://doi.org/10.3354/meps12863

ABSTRACT: The effects of selecting system compartments in the calculation of Ecological Network Analysis (ENA) indices were studied based on data collected in the 3 main reaches of the Ems estuary. For each reach, ENA was applied to (1) a set of carbon flow models in which only the living compartments were hand-balanced, and (2) a set in which living and non-living compartments were hand-balanced. The models considered represent a full food web at the highest species resolution level and 4 food web subsets representing benthic macrofauna, benthic macrofauna plus demersal fish and epifauna, all fish, and all birds. Each of the 30 models consist of 11 to 57 compartments (species, functional groups, C-pools). Results demonstrate that the food web subsets are predominantly responsible for the variation in the ENA indices (relative ascendency: 15% of its maximum of 1; internal relative ascendency: 21%). The use of food web subsets also leads to increased variation in the Finn cycling index, effective link density, trophic depth and robustness. The use of subsets is therefore discouraged. The added value of robustness as a practical index was explored in relation to ascendency-related (i.e. information-related) indices, but its use could not be supported because its range was too limited. The results also indicate that incomplete food webs significantly influence the size of the ‘window of vitality’. ENA works well when food web subsets include the full trophic biomass pyramid, but top-down studies focussing on so-called iconic species are not advisable for assessments. The closer to the full food web, the better the results of ENA correspond with those of a full food web.

KEY WORDS: Assessment · Ecosystem · Indicators · Carbon flows · enaR · ENA indices · Metrics