Stoichiometric multitrophic networks reveal significance of land-sea interaction to ecosystem function in a subtropical nutrient-poor bight, South Africa.

Nearshore marine ecosystems can benefit from their interaction with adjacent ecosystems, especially if they alleviate nutrient limitations in nutrient poor areas. This was the case in our oligo- to mesotrophic study area, the KwaZulu-Natal Bight on the South African subtropical east coast, which is bordered by the Agulhas current. We built stoichiometric, multitrophic ecosystem networks depicting biomass and material flows of carbon, nitrogen and phosphorus in three subsystems of the bight. The networks were analysed to investigate whether the southern, middle and northern bight function similarly in terms of their productivity, transfer efficiency between trophic levels, material cycling, and nutrient limitations. The middle region of the bight was clearly influenced by nutrient additions from the Thukela River, as it had the highest ecosystem productivity, lower transfer efficiencies and degree of cycling. Most nodes in the networks were limited by phosphorus, followed by nitrogen. The middle region adjacent to the Thukela River showed a lower proportion of P limitation especially in summer. Interestingly, there was a clear distinction in sensitivities to nutrient limitations between lower and higher trophic level organisms. This is a reflection of their discrepant nutrient turnover times that are either higher, or lower, than that of the systems, and which might provide a balance to the system through this antagonistic influence. Furthermore, by tracking the stoichiometry through entire food webs it appeared how important the role of lower trophic level organisms was to regulate stoichiometry to more suitable ratios for higher trophic level requirements. Although we gained good insight into the behaviour of the three subsystems in the KZN Bight and the role of terrestrial influence on their functioning, a merged approach of incorporating data on metabolic constraints derived from experiments could further improve the representativeness of multitrophic stoichiometric ecosystem networks